bookloupe-testing: bookloupe/bookloupe.c@865063352146 (annotated)

ali@0	1	/*************************************************************************/
ali@40	2	/* bookloupe--check for assorted weirdnesses in a PG candidate text file */
ali@0	3	/* */
ali@0	4	/* Copyright 2000-2005 Jim Tinsley <jtinsley@pobox.com> */
ali@40	5	/* Copyright 2012- J. Ali Harlow <ali@juiblex.co.uk> */
ali@0	6	/* */
ali@0	7	/* This program is free software; you can redistribute it and/or modify */
ali@0	8	/* it under the terms of the GNU General Public License as published by */
ali@0	9	/* the Free Software Foundation; either version 2 of the License, or */
ali@0	10	/* (at your option) any later version. */
ali@0	11	/* */
ali@0	12	/* This program is distributed in the hope that it will be useful, */
ali@0	13	/* but WITHOUT ANY WARRANTY; without even the implied warranty of */
ali@40	14	/* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the */
ali@0	15	/* GNU General Public License for more details. */
ali@0	16	/* */
ali@0	17	/* You should have received a copy of the GNU General Public License */
ali@40	18	/* along with this program. If not, see <http://www.gnu.org/licenses/>. */
ali@0	19	/*************************************************************************/
ali@0	20
ali@0	21	#include <stdio.h>
ali@0	22	#include <stdlib.h>
ali@0	23	#include <string.h>
ali@0	24	#include <ctype.h>
ali@0	25
ali@0	26	#define MAXWORDLEN 80 /* max length of one word */
ali@0	27	#define LINEBUFSIZE 2048 /* buffer size for an input line */
ali@0	28
ali@0	29	#define MAX_USER_TYPOS 1000
ali@0	30	#define USERTYPO_FILE "gutcheck.typ"
ali@0	31
ali@0	32	#ifndef MAX_PATH
ali@0	33	#define MAX_PATH 16384
ali@0	34	#endif
ali@0	35
ali@0	36	char aline[LINEBUFSIZE];
ali@0	37	char prevline[LINEBUFSIZE];
ali@0	38
ali@40	39	/* Common typos. */
ali@40	40	char *typo[] = {
ali@40	41	"teh", "th", "og", "fi", "ro", "adn", "yuo", "ot", "fo", "thet", "ane",
ali@40	42	"nad", "te", "ig", "acn", "ahve", "alot", "anbd", "andt", "awya", "aywa",
ali@40	43	"bakc", "om", "btu", "byt", "cna", "cxan", "coudl", "dont", "didnt",
ali@40	44	"couldnt", "wouldnt", "doesnt", "shouldnt", "doign", "ehr", "hmi", "hse",
ali@40	45	"esle", "eyt", "fitrs", "firts", "foudn", "frmo", "fromt", "fwe", "gaurd",
ali@40	46	"gerat", "goign", "gruop", "haev", "hda", "hearign", "seeign", "sayign",
ali@40	47	"herat", "hge", "hsa", "hsi", "hte", "htere", "htese", "htey", "htis",
ali@40	48	"hvae", "hwich", "idae", "ihs", "iits", "int", "iwll", "iwth", "jsut",
ali@40	49	"loev", "sefl", "myu", "nkow", "nver", "nwe", "nwo", "ocur", "ohter",
ali@40	50	"omre", "onyl", "otehr", "otu", "owrk", "owuld", "peice", "peices",
ali@40	51	"peolpe", "peopel", "perhasp", "perhpas", "pleasent", "poeple", "porblem",
ali@40	52	"porblems", "rwite", "saidt", "saidh", "saids", "seh", "smae", "smoe",
ali@40	53	"sohw", "stnad", "stopry", "stoyr", "stpo", "tahn", "taht", "tath",
ali@40	54	"tehy", "tghe", "tghis", "theri", "theyll", "thgat", "thge", "thier",
ali@40	55	"thna", "thne", "thnig", "thnigs", "thsi", "thsoe", "thta", "timne",
ali@40	56	"tirne", "tkae", "tthe", "tyhat", "tyhe", "veyr", "vou", "vour", "vrey",
ali@40	57	"waht", "wasnt", "awtn", "watn", "wehn", "whic", "whcih", "whihc", "whta",
ali@40	58	"wihch", "wief", "wiht", "witha", "wiull", "wnat", "wnated", "wnats",
ali@40	59	"woh", "wohle", "wokr", "woudl", "wriet", "wrod", "wroet", "wroking",
ali@40	60	"wtih", "wuould", "wya", "yera", "yeras", "yersa", "yoiu", "youve",
ali@40	61	"ytou", "yuor", "abead", "ahle", "ahout", "ahove", "altbough", "balf",
ali@40	62	"bardly", "bas", "bave", "baving", "bebind", "beld", "belp", "belped",
ali@40	63	"ber", "bere", "bim", "bis", "bome", "bouse", "bowever", "buge",
ali@40	64	"dehates", "deht", "han", "hecause", "hecome", "heen", "hefore", "hegan",
ali@40	65	"hegin", "heing", "helieve", "henefit", "hetter", "hetween", "heyond",
ali@40	66	"hig", "higber", "huild", "huy", "hy", "jobn", "joh", "meanwbile",
ali@40	67	"memher", "memhers", "numher", "numhers", "perbaps", "prohlem", "puhlic",
ali@40	68	"witbout", "arn", "hin", "hirn", "wrok", "wroked", "amd", "aud",
ali@40	69	"prornise", "prornised", "modem", "bo", "heside", "chapteb", "chaptee",
ali@40	70	"se", ""
ali@40	71	};
ali@0	72
ali@0	73	char *usertypo[MAX_USER_TYPOS];
ali@0	74
ali@40	75	/* Common abbreviations and other OK words not to query as typos. */
ali@40	76	char *okword[] = {
ali@40	77	"mr", "mrs", "mss", "mssrs", "ft", "pm", "st", "dr", "hmm", "h'm", "hmmm",
ali@40	78	"rd", "sh", "br", "pp", "hm", "cf", "jr", "sr", "vs", "lb", "lbs", "ltd",
ali@40	79	"pompeii","hawaii","hawaiian", "hotbed", "heartbeat", "heartbeats",
ali@40	80	"outbid", "outbids", "frostbite", "frostbitten", ""
ali@40	81	};
ali@0	82
ali@40	83	/* Common abbreviations that cause otherwise unexplained periods. */
ali@40	84	char *abbrev[] = {
ali@40	85	"cent", "cents", "viz", "vol", "vols", "vid", "ed", "al", "etc", "op",
ali@40	86	"cit", "deg", "min", "chap", "oz", "mme", "mlle", "mssrs", ""
ali@40	87	};
ali@0	88
ali@40	89	/*
ali@40	90	* Two-Letter combinations that rarely if ever start words,
ali@40	91	* but are common scannos or otherwise common letter combinations.
ali@40	92	*/
ali@40	93	char *nostart[] = {
ali@40	94	"hr", "hl", "cb", "sb", "tb", "wb", "tl", "tn", "rn", "lt", "tj", ""
ali@40	95	};
ali@0	96
ali@40	97	/*
ali@40	98	* Two-Letter combinations that rarely if ever end words,
ali@40	99	* but are common scannos or otherwise common letter combinations.
ali@40	100	*/
ali@40	101	char *noend[] = {
ali@40	102	"cb", "gb", "pb", "sb", "tb", "wh", "fr", "br", "qu", "tw", "gl", "fl",
ali@40	103	"sw", "gr", "sl", "cl", "iy", ""
ali@40	104	};
ali@0	105
ali@40	106	char *markup[] = {
ali@40	107	"a", "b", "big", "blockquote", "body", "br", "center", "col", "div", "em",
ali@40	108	"font", "h1", "h2", "h3", "h4", "h5", "h6", "head", "hr", "html", "i",
ali@40	109	"img", "li", "meta", "ol", "p", "pre", "small", "span", "strong", "sub",
ali@40	110	"sup", "table", "td", "tfoot", "thead", "title", "tr", "tt", "u", "ul", ""
ali@40	111	};
ali@0	112
ali@40	113	char *DPmarkup[] = {
ali@40	114	"<sc>", "</sc>", "/", "/", "/#", "#/", "/$", "$/", "<tb>", ""
ali@40	115	};
ali@0	116
ali@40	117	char *nocomma[] = {
ali@40	118	"the", "it's", "their", "an", "mrs", "a", "our", "that's", "its", "whose",
ali@40	119	"every", "i'll", "your", "my", "mr", "mrs", "mss", "mssrs", "ft", "pm",
ali@40	120	"st", "dr", "rd", "pp", "cf", "jr", "sr", "vs", "lb", "lbs", "ltd", "i'm",
ali@40	121	"during", "let", "toward", "among", ""
ali@40	122	};
ali@0	123
ali@40	124	char *noperiod[] = {
ali@40	125	"every", "i'm", "during", "that's", "their", "your", "our", "my", "or",
ali@40	126	"and", "but", "as", "if", "the", "its", "it's", "until", "than", "whether",
ali@40	127	"i'll", "whose", "who", "because", "when", "let", "till", "very", "an",
ali@40	128	"among", "those", "into", "whom", "having", "thence", ""
ali@40	129	};
ali@0	130
ali@40	131	char vowels[] = "aeiouàáâãäæèéêëìíîïòóôõöùúûü";
ali@0	132
ali@0	133	struct {
ali@0	134	char *htmlent;
ali@0	135	char *htmlnum;
ali@0	136	char *textent;
ali@40	137	} entities[] = {
ali@40	138	"&", "&", "&",
ali@40	139	"<", "<", "<",
ali@40	140	">", ">", ">",
ali@40	141	"°", "°", " degrees",
ali@40	142	"£", "£", "L",
ali@40	143	""", """, "\"", /* quotation mark = APL quote */
ali@40	144	"&OElig;", "Œ", "OE", /* latin capital ligature OE */
ali@40	145	"&oelig;", "œ", "oe", /* latin small ligature oe */
ali@40	146	"&Scaron;", "Š", "S", /* latin capital letter S with caron */
ali@40	147	"&scaron;", "š", "s", /* latin small letter s with caron */
ali@40	148	"&Yuml;", "Ÿ", "Y", /* latin capital letter Y with diaeresis */
ali@40	149	"&circ;", "ˆ", "", /* modifier letter circumflex accent */
ali@40	150	"&tilde;", "˜", "~", /* small tilde, U+02DC ISOdia */
ali@40	151	"&ensp;", " ", " ", /* en space, U+2002 ISOpub */
ali@40	152	"&emsp;", " ", " ", /* em space, U+2003 ISOpub */
ali@40	153	" ", " ", " ", /* thin space, U+2009 ISOpub */
ali@40	154	"–", "–", "-", /* en dash, U+2013 ISOpub */
ali@40	155	"—", "—", "--", /* em dash, U+2014 ISOpub */
ali@40	156	"’", "’", "'", /* right single quotation mark */
ali@40	157	"&sbquo;", "‚", "'", /* single low-9 quotation mark */
ali@40	158	"“", "“", "\"", /* left double quotation mark */
ali@40	159	"”", "”", "\"", /* right double quotation mark */
ali@40	160	"&bdquo;", "„", "\"", /* double low-9 quotation mark */
ali@40	161	"&lsaquo;", "‹", "\"", /* single left-pointing angle quotation mark */
ali@40	162	"&rsaquo;", "›", "\"", /* single right-pointing angle quotation mark */
ali@40	163	" ", " ", " ", /* no-break space = non-breaking space, */
ali@40	164	"¡", "¡", "!", /* inverted exclamation mark */
ali@40	165	"¢", "¢", "c", /* cent sign */
ali@40	166	"£", "£", "L", /* pound sign */
ali@40	167	"¤", "¤", "$", /* currency sign */
ali@40	168	"¥", "¥", "Y", /* yen sign = yuan sign */
ali@40	169	"§", "§", "--", /* section sign */
ali@40	170	"¨", "¨", " ", /* diaeresis = spacing diaeresis */
ali@40	171	"©", "©", "(C) ", /* copyright sign */
ali@40	172	"ª", "ª", " ", /* feminine ordinal indicator */
ali@40	173	"«", "«", "\"", /* left-pointing double angle quotation mark */
ali@40	174	"", "", "-", /* soft hyphen = discretionary hyphen */
ali@40	175	"®", "®", "(R) ", /* registered sign = registered trade mark sign */
ali@40	176	"¯", "¯", " ", /* macron = spacing macron = overline */
ali@40	177	"°", "°", " degrees", /* degree sign */
ali@40	178	"±", "±", "+-", /* plus-minus sign = plus-or-minus sign */
ali@40	179	"²", "²", "2", /* superscript two = superscript digit two */
ali@40	180	"³", "³", "3", /* superscript three = superscript digit three */
ali@40	181	"´", "´", " ", /* acute accent = spacing acute */
ali@40	182	"µ", "µ", "m", /* micro sign */
ali@40	183	"¶", "¶", "--", /* pilcrow sign = paragraph sign */
ali@40	184	"¸", "¸", " ", /* cedilla = spacing cedilla */
ali@40	185	"¹", "¹", "1", /* superscript one = superscript digit one */
ali@40	186	"º", "º", " ", /* masculine ordinal indicator */
ali@40	187	"»", "»", "\"", /* right-pointing double angle quotation mark */
ali@40	188	"¼", "¼", "1/4", /* vulgar fraction one quarter */
ali@40	189	"½", "½", "1/2", /* vulgar fraction one half */
ali@40	190	"¾", "¾", "3/4", /* vulgar fraction three quarters */
ali@40	191	"¿", "¿", "?", /* inverted question mark */
ali@40	192	"À", "À", "A", /* latin capital letter A with grave */
ali@40	193	"Á", "Á", "A", /* latin capital letter A with acute */
ali@40	194	"Â", "Â", "A", /* latin capital letter A with circumflex */
ali@40	195	"Ã", "Ã", "A", /* latin capital letter A with tilde */
ali@40	196	"Ä", "Ä", "A", /* latin capital letter A with diaeresis */
ali@40	197	"Å", "Å", "A", /* latin capital letter A with ring above */
ali@40	198	"Æ", "Æ", "AE", /* latin capital letter AE */
ali@40	199	"Ç", "Ç", "C", /* latin capital letter C with cedilla */
ali@40	200	"È", "È", "E", /* latin capital letter E with grave */
ali@40	201	"É", "É", "E", /* latin capital letter E with acute */
ali@40	202	"Ê", "Ê", "E", /* latin capital letter E with circumflex */
ali@40	203	"Ë", "Ë", "E", /* latin capital letter E with diaeresis */
ali@40	204	"Ì", "Ì", "I", /* latin capital letter I with grave */
ali@40	205	"Í", "Í", "I", /* latin capital letter I with acute */
ali@40	206	"Î", "Î", "I", /* latin capital letter I with circumflex */
ali@40	207	"Ï", "Ï", "I", /* latin capital letter I with diaeresis */
ali@40	208	"Ð", "Ð", "E", /* latin capital letter ETH */
ali@40	209	"Ñ", "Ñ", "N", /* latin capital letter N with tilde */
ali@40	210	"Ò", "Ò", "O", /* latin capital letter O with grave */
ali@40	211	"Ó", "Ó", "O", /* latin capital letter O with acute */
ali@40	212	"Ô", "Ô", "O", /* latin capital letter O with circumflex */
ali@40	213	"Õ", "Õ", "O", /* latin capital letter O with tilde */
ali@40	214	"Ö", "Ö", "O", /* latin capital letter O with diaeresis */
ali@40	215	"×", "×", "", / multiplication sign */
ali@40	216	"Ø", "Ø", "O", /* latin capital letter O with stroke */
ali@40	217	"Ù", "Ù", "U", /* latin capital letter U with grave */
ali@40	218	"Ú", "Ú", "U", /* latin capital letter U with acute */
ali@40	219	"Û", "Û", "U", /* latin capital letter U with circumflex */
ali@40	220	"Ü", "Ü", "U", /* latin capital letter U with diaeresis */
ali@40	221	"Ý", "Ý", "Y", /* latin capital letter Y with acute */
ali@40	222	"Þ", "Þ", "TH", /* latin capital letter THORN */
ali@40	223	"ß", "ß", "sz", /* latin small letter sharp s = ess-zed */
ali@40	224	"à", "à", "a", /* latin small letter a with grave */
ali@40	225	"á", "á", "a", /* latin small letter a with acute */
ali@40	226	"â", "â", "a", /* latin small letter a with circumflex */
ali@40	227	"ã", "ã", "a", /* latin small letter a with tilde */
ali@40	228	"ä", "ä", "a", /* latin small letter a with diaeresis */
ali@40	229	"å", "å", "a", /* latin small letter a with ring above */
ali@40	230	"æ", "æ", "ae", /* latin small letter ae */
ali@40	231	"ç", "ç", "c", /* latin small letter c with cedilla */
ali@40	232	"è", "è", "e", /* latin small letter e with grave */
ali@40	233	"é", "é", "e", /* latin small letter e with acute */
ali@40	234	"ê", "ê", "e", /* latin small letter e with circumflex */
ali@40	235	"ë", "ë", "e", /* latin small letter e with diaeresis */
ali@40	236	"ì", "ì", "i", /* latin small letter i with grave */
ali@40	237	"í", "í", "i", /* latin small letter i with acute */
ali@40	238	"î", "î", "i", /* latin small letter i with circumflex */
ali@40	239	"ï", "ï", "i", /* latin small letter i with diaeresis */
ali@40	240	"ð", "ð", "eth", /* latin small letter eth */
ali@40	241	"ñ", "ñ", "n", /* latin small letter n with tilde */
ali@40	242	"ò", "ò", "o", /* latin small letter o with grave */
ali@40	243	"ó", "ó", "o", /* latin small letter o with acute */
ali@40	244	"ô", "ô", "o", /* latin small letter o with circumflex */
ali@40	245	"õ", "õ", "o", /* latin small letter o with tilde */
ali@40	246	"ö", "ö", "o", /* latin small letter o with diaeresis */
ali@40	247	"÷", "÷", "/", /* division sign */
ali@40	248	"ø", "ø", "o", /* latin small letter o with stroke */
ali@40	249	"ù", "ù", "u", /* latin small letter u with grave */
ali@40	250	"ú", "ú", "u", /* latin small letter u with acute */
ali@40	251	"û", "û", "u", /* latin small letter u with circumflex */
ali@40	252	"ü", "ü", "u", /* latin small letter u with diaeresis */
ali@40	253	"ý", "ý", "y", /* latin small letter y with acute */
ali@40	254	"þ", "þ", "th", /* latin small letter thorn */
ali@40	255	"ÿ", "ÿ", "y", /* latin small letter y with diaeresis */
ali@40	256	"", ""
ali@40	257	};
ali@40	258
ali@40	259	/* special characters */
ali@0	260	#define CHAR_SPACE 32
ali@0	261	#define CHAR_TAB 9
ali@0	262	#define CHAR_LF 10
ali@0	263	#define CHAR_CR 13
ali@0	264	#define CHAR_DQUOTE 34
ali@0	265	#define CHAR_SQUOTE 39
ali@0	266	#define CHAR_OPEN_SQUOTE 96
ali@0	267	#define CHAR_TILDE 126
ali@0	268	#define CHAR_ASTERISK 42
ali@0	269	#define CHAR_FORESLASH 47
ali@0	270	#define CHAR_CARAT 94
ali@0	271
ali@0	272	#define CHAR_UNDERSCORE '_'
ali@0	273	#define CHAR_OPEN_CBRACK '{'
ali@0	274	#define CHAR_CLOSE_CBRACK '}'
ali@0	275	#define CHAR_OPEN_RBRACK '('
ali@0	276	#define CHAR_CLOSE_RBRACK ')'
ali@0	277	#define CHAR_OPEN_SBRACK '['
ali@0	278	#define CHAR_CLOSE_SBRACK ']'
ali@0	279
ali@40	280	/* longest and shortest normal PG line lengths */
ali@0	281	#define LONGEST_PG_LINE 75
ali@0	282	#define WAY_TOO_LONG 80
ali@0	283	#define SHORTEST_PG_LINE 55
ali@0	284
ali@0	285	#define SWITCHES "ESTPXLOYHWVMUD" /* switches:- */
ali@0	286	/* D - ignore DP-specific markup */
ali@0	287	/* E - echo queried line */
ali@0	288	/* S - check single quotes */
ali@0	289	/* T - check common typos */
ali@0	290	/* P - require closure of quotes on */
ali@0	291	/* every paragraph */
ali@0	292	/* X - "Trust no one" :-) Paranoid! */
ali@0	293	/* Queries everything */
ali@0	294	/* L - line end checking defaults on */
ali@0	295	/* -L turns it off */
ali@0	296	/* O - overview. Just shows counts. */
ali@0	297	/* Y - puts errors to stdout */
ali@0	298	/* instead of stderr */
ali@0	299	/* H - Echoes header fields */
ali@0	300	/* M - Ignore markup in < > */
ali@0	301	/* U - Use file of User-defined Typos*/
ali@0	302	/* W - Defaults for use on Web upload*/
ali@0	303	/* V - Verbose - list EVERYTHING! */
ali@0	304	#define SWITNO 14 /* max number of switch parms */
ali@0	305	/* - used for defining array-size */
ali@0	306	#define MINARGS 1 /* minimum no of args excl switches */
ali@0	307	#define MAXARGS 1 /* maximum no of args excl switches */
ali@0	308
ali@0	309	int pswit[SWITNO]; /* program switches set by SWITCHES */
ali@0	310
ali@0	311	#define ECHO_SWITCH 0
ali@0	312	#define SQUOTE_SWITCH 1
ali@0	313	#define TYPO_SWITCH 2
ali@0	314	#define QPARA_SWITCH 3
ali@0	315	#define PARANOID_SWITCH 4
ali@0	316	#define LINE_END_SWITCH 5
ali@0	317	#define OVERVIEW_SWITCH 6
ali@0	318	#define STDOUT_SWITCH 7
ali@0	319	#define HEADER_SWITCH 8
ali@0	320	#define WEB_SWITCH 9
ali@0	321	#define VERBOSE_SWITCH 10
ali@0	322	#define MARKUP_SWITCH 11
ali@0	323	#define USERTYPO_SWITCH 12
ali@0	324	#define DP_SWITCH 13
ali@0	325
ali@0	326	long cnt_dquot; /* for overview mode, count of doublequote queries */
ali@0	327	long cnt_squot; /* for overview mode, count of singlequote queries */
ali@0	328	long cnt_brack; /* for overview mode, count of brackets queries */
ali@0	329	long cnt_bin; /* for overview mode, count of non-ASCII queries */
ali@0	330	long cnt_odd; /* for overview mode, count of odd character queries */
ali@0	331	long cnt_long; /* for overview mode, count of long line errors */
ali@0	332	long cnt_short; /* for overview mode, count of short line queries */
ali@0	333	long cnt_punct; /* for overview mode, count of punctuation and spacing queries */
ali@0	334	long cnt_dash; /* for overview mode, count of dash-related queries */
ali@0	335	long cnt_word; /* for overview mode, count of word queries */
ali@0	336	long cnt_html; /* for overview mode, count of html queries */
ali@0	337	long cnt_lineend; /* for overview mode, count of line-end queries */
ali@40	338	long cnt_spacend; /* count of lines with space at end */
ali@0	339	long linecnt; /* count of total lines in the file */
ali@40	340	long checked_linecnt; /* count of lines actually checked */
ali@0	341
ali@0	342	void proghelp(void);
ali@0	343	void procfile(char *);
ali@0	344
ali@0	345	#define LOW_THRESHOLD 0
ali@0	346	#define HIGH_THRESHOLD 1
ali@0	347
ali@0	348	#define START 0
ali@0	349	#define END 1
ali@0	350	#define PREV 0
ali@0	351	#define NEXT 1
ali@0	352	#define FIRST_OF_PAIR 0
ali@0	353	#define SECOND_OF_PAIR 1
ali@0	354
ali@0	355	#define MAX_WORDPAIR 1000
ali@0	356
ali@0	357	char running_from[MAX_PATH];
ali@0	358
ali@0	359	int mixdigit(char *);
ali@54	360	const char getaword(const char ,char *);
ali@40	361	int matchword(char ,char );
ali@40	362	char flgets(char ,int,FILE *,long);
ali@0	363	void lowerit(char *);
ali@0	364	int gcisalpha(unsigned char);
ali@0	365	int gcisdigit(unsigned char);
ali@0	366	int gcisletter(unsigned char);
ali@40	367	char gcstrchr(char s,char c);
ali@0	368	void postprocess_for_HTML(char *);
ali@0	369	char linehasmarkup(char );
ali@0	370	char losemarkup(char );
ali@40	371	int tagcomp(char ,char );
ali@0	372	char loseentities(char );
ali@0	373	int isroman(char *);
ali@0	374	int usertypo_count;
ali@0	375	void postprocess_for_DP(char *);
ali@0	376
ali@0	377	char wrk[LINEBUFSIZE];
ali@0	378
ali@40	379	#define MAX_QWORD 50
ali@40	380	#define MAX_QWORD_LENGTH 40
ali@0	381	char qword[MAX_QWORD][MAX_QWORD_LENGTH];
ali@0	382	signed int dupcnt[MAX_QWORD];
ali@0	383
ali@40	384	int main(int argc,char **argv)
ali@0	385	{
ali@40	386	char argsw,s;
ali@40	387	int i,switno,invarg;
ali@0	388	char usertypo_file[MAX_PATH];
ali@0	389	FILE *usertypofile;
ali@40	390	if (strlen(argv[0])<sizeof(running_from))
ali@40	391	/* save the path to the executable */
ali@40	392	strcpy(running_from,argv[0]);
ali@0	393	/* find out what directory we're running from */
ali@40	394	s=running_from+strlen(running_from);
ali@40	395	for (;s!='/' && s!='\\' && s>=running_from;s--)
ali@40	396	*s=0;
ali@40	397	switno=strlen(SWITCHES);
ali@40	398	for (i=switno;--i>0;)
ali@40	399	pswit[i]=0; /* initialise switches */
ali@40	400	/*
ali@40	401	* Standard loop to extract switches.
ali@40	402	* When we come out of this loop, the arguments will be
ali@40	403	* in argv[0] upwards and the switches used will be
ali@40	404	* represented by their equivalent elements in pswit[]
ali@40	405	*/
ali@40	406	while (--argc>0 && **++argv=='-')
ali@40	407	for (argsw=argv[0]+1;*argsw!='\0';argsw++)
ali@40	408	for (i=switno,invarg=1;(--i>=0) && invarg==1;)
ali@40	409	if ((toupper(*argsw))==SWITCHES[i])
ali@40	410	{
ali@40	411	invarg=0;
ali@40	412	pswit[i]=1;
ali@40	413	}
ali@40	414	/* Paranoid checking is turned OFF, not on, by its switch */
ali@40	415	pswit[PARANOID_SWITCH]^=1;
ali@40	416	if (pswit[PARANOID_SWITCH])
ali@40	417	/* if running in paranoid mode force typo checks as well */
ali@40	418	pswit[TYPO_SWITCH]=pswit[TYPO_SWITCH]^1;
ali@40	419	/* Line-end checking is turned OFF, not on, by its switch */
ali@40	420	pswit[LINE_END_SWITCH]^=1;
ali@40	421	/* Echoing is turned OFF, not on, by its switch */
ali@40	422	pswit[ECHO_SWITCH]^=1;
ali@40	423	if (pswit[OVERVIEW_SWITCH])
ali@40	424	/* just print summary; don't echo */
ali@40	425	pswit[ECHO_SWITCH]=0;
ali@40	426	/*
ali@40	427	* Web uploads - for the moment, this is really just a placeholder
ali@40	428	* until we decide what processing we really want to do on web uploads
ali@40	429	*/
ali@40	430	if (pswit[WEB_SWITCH])
ali@40	431	{
ali@40	432	/* specific override for web uploads */
ali@40	433	pswit[ECHO_SWITCH]=1;
ali@40	434	pswit[SQUOTE_SWITCH]=0;
ali@40	435	pswit[TYPO_SWITCH]=1;
ali@40	436	pswit[QPARA_SWITCH]=0;
ali@40	437	pswit[PARANOID_SWITCH]=1;
ali@40	438	pswit[LINE_END_SWITCH]=0;
ali@40	439	pswit[OVERVIEW_SWITCH]=0;
ali@40	440	pswit[STDOUT_SWITCH]=0;
ali@40	441	pswit[HEADER_SWITCH]=1;
ali@40	442	pswit[VERBOSE_SWITCH]=0;
ali@40	443	pswit[MARKUP_SWITCH]=0;
ali@40	444	pswit[USERTYPO_SWITCH]=0;
ali@40	445	pswit[DP_SWITCH]=0;
ali@40	446	}
ali@40	447	if (argc<MINARGS \|\| argc>MAXARGS)
ali@40	448	{
ali@40	449	/* check number of args */
ali@0	450	proghelp();
ali@40	451	return 1;
ali@40	452	}
ali@0	453	/* read in the user-defined stealth scanno list */
ali@40	454	if (pswit[USERTYPO_SWITCH])
ali@40	455	{
ali@40	456	/* ... we were told we had one! */
ali@40	457	usertypofile=fopen(USERTYPO_FILE,"rb");
ali@40	458	if (!usertypofile)
ali@40	459	{
ali@40	460	/* not in cwd. try excuteable directory. */
ali@40	461	strcpy(usertypo_file,running_from);
ali@40	462	strcat(usertypo_file,USERTYPO_FILE);
ali@40	463	usertypofile=fopen(usertypo_file,"rb");
ali@40	464	if (!usertypofile) {
ali@40	465	/* we ain't got no user typo file! */
ali@40	466	printf(" --> I couldn't find gutcheck.typ "
ali@40	467	"-- proceeding without user typos.\n");
ali@40	468	}
ali@40	469	}
ali@40	470	usertypo_count=0;
ali@40	471	if (usertypofile)
ali@40	472	{
ali@40	473	/* we managed to open a User Typo File! */
ali@40	474	if (pswit[USERTYPO_SWITCH])
ali@40	475	{
ali@40	476	while (flgets(aline,LINEBUFSIZE-1,usertypofile,
ali@40	477	(long)usertypo_count))
ali@40	478	{
ali@40	479	if (strlen(aline)>1)
ali@40	480	{
ali@40	481	if ((int)*aline>33)
ali@40	482	{
ali@40	483	s=malloc(strlen(aline)+1);
ali@40	484	if (!s)
ali@40	485	{
ali@40	486	fprintf(stderr,"bookloupe: cannot get enough "
ali@40	487	"memory for user typo file!\n");
ali@0	488	exit(1);
ali@40	489	}
ali@40	490	strcpy(s,aline);
ali@40	491	usertypo[usertypo_count]=s;
ali@0	492	usertypo_count++;
ali@40	493	if (usertypo_count>=MAX_USER_TYPOS)
ali@40	494	{
ali@40	495	printf(" --> Only %d user-defined typos "
ali@42	496	"allowed: ignoring the rest\n",
ali@42	497	MAX_USER_TYPOS);
ali@0	498	break;
ali@40	499	}
ali@40	500	}
ali@40	501	}
ali@40	502	}
ali@40	503	}
ali@0	504	fclose(usertypofile);
ali@40	505	}
ali@40	506	}
ali@40	507	fprintf(stderr,"bookloupe: Check and report on an e-text\n");
ali@40	508	cnt_dquot=cnt_squot=cnt_brack=cnt_bin=cnt_odd=cnt_long=
ali@40	509	cnt_short=cnt_punct=cnt_dash=cnt_word=cnt_html=cnt_lineend=
ali@40	510	cnt_spacend=0;
ali@0	511	procfile(argv[0]);
ali@40	512	if (pswit[OVERVIEW_SWITCH])
ali@40	513	{
ali@40	514	printf(" Checked %ld lines of %ld (head+foot = %ld)\n\n",
ali@40	515	checked_linecnt,linecnt,linecnt-checked_linecnt);
ali@40	516	printf(" --------------- Queries found --------------\n");
ali@40	517	if (cnt_long)
ali@40	518	printf(" Long lines: %14ld\n",cnt_long);
ali@40	519	if (cnt_short)
ali@40	520	printf(" Short lines: %14ld\n",cnt_short);
ali@40	521	if (cnt_lineend)
ali@40	522	printf(" Line-end problems: %14ld\n",cnt_lineend);
ali@40	523	if (cnt_word)
ali@40	524	printf(" Common typos: %14ld\n",cnt_word);
ali@40	525	if (cnt_dquot)
ali@40	526	printf(" Unmatched quotes: %14ld\n",cnt_dquot);
ali@40	527	if (cnt_squot)
ali@40	528	printf(" Unmatched SingleQuotes: %14ld\n",cnt_squot);
ali@40	529	if (cnt_brack)
ali@40	530	printf(" Unmatched brackets: %14ld\n",cnt_brack);
ali@40	531	if (cnt_bin)
ali@40	532	printf(" Non-ASCII characters: %14ld\n",cnt_bin);
ali@40	533	if (cnt_odd)
ali@40	534	printf(" Proofing characters: %14ld\n",cnt_odd);
ali@40	535	if (cnt_punct)
ali@40	536	printf(" Punctuation & spacing queries: %14ld\n",cnt_punct);
ali@40	537	if (cnt_dash)
ali@40	538	printf(" Non-standard dashes: %14ld\n",cnt_dash);
ali@40	539	if (cnt_html)
ali@40	540	printf(" Possible HTML tags: %14ld\n",cnt_html);
ali@0	541	printf("\n");
ali@40	542	printf(" TOTAL QUERIES %14ld\n",
ali@40	543	cnt_dquot+cnt_squot+cnt_brack+cnt_bin+cnt_odd+cnt_long+
ali@40	544	cnt_short+cnt_punct+cnt_dash+cnt_word+cnt_html+cnt_lineend);
ali@40	545	}
ali@40	546	return 0;
ali@0	547	}
ali@0	548
ali@41	549	struct first_pass_results {
ali@41	550	long firstline,astline;
ali@41	551	long footerline,totlen,binlen,alphalen,endquote_count,shortline,dotcomma;
ali@41	552	long fslashline,hyphens,longline,verylongline,htmcount,standalone_digit;
ali@41	553	long spacedash,emdash,space_emdash,non_PG_space_emdash,PG_space_emdash;
ali@41	554	signed int Dutchcount,Frenchcount;
ali@41	555	};
ali@41	556
ali@40	557	/*
ali@41	558	* first_pass:
ali@40	559	*
ali@41	560	* Run a first pass - verify that it's a valid PG
ali@41	561	* file, decide whether to report some things that
ali@41	562	* occur many times in the text like long or short
ali@41	563	* lines, non-standard dashes, etc.
ali@40	564	*/
ali@41	565	struct first_pass_results first_pass(FILE infile)
ali@0	566	{
ali@54	567	char laststart=CHAR_SPACE;
ali@54	568	const char *s;
ali@41	569	signed int i,llen;
ali@41	570	unsigned int lastlen=0,lastblen=0;
ali@41	571	long spline=0,nspline=0;
ali@41	572	static struct first_pass_results results={0};
ali@41	573	char inword[MAXWORDLEN]="";
ali@40	574	while (fgets(aline,LINEBUFSIZE-1,infile))
ali@40	575	{
ali@40	576	while (aline[strlen(aline)-1]==10 \|\| aline[strlen(aline)-1]==13)
ali@40	577	aline[strlen(aline)-1]=0;
ali@0	578	linecnt++;
ali@40	579	if (strstr(aline,"*END") && strstr(aline,"SMALL PRINT") &&
ali@40	580	(strstr(aline,"PUBLIC DOMAIN") \|\| strstr(aline,"COPYRIGHT")))
ali@40	581	{
ali@0	582	if (spline)
ali@0	583	printf(" --> Duplicate header?\n");
ali@40	584	spline=linecnt+1; /* first line of non-header text, that is */
ali@40	585	}
ali@40	586	if (!strncmp(aline,"*** START",9) && strstr(aline,"PROJECT GUTENBERG"))
ali@40	587	{
ali@0	588	if (nspline)
ali@0	589	printf(" --> Duplicate header?\n");
ali@40	590	nspline=linecnt+1; /* first line of non-header text, that is */
ali@40	591	}
ali@40	592	if (spline \|\| nspline)
ali@40	593	{
ali@0	594	lowerit(aline);
ali@40	595	if (strstr(aline,"end") && strstr(aline,"project gutenberg"))
ali@40	596	{
ali@40	597	if (strstr(aline,"end")<strstr(aline,"project gutenberg"))
ali@40	598	{
ali@41	599	if (results.footerline)
ali@40	600	{
ali@40	601	/* it's an old-form header - we can detect duplicates */
ali@40	602	if (!nspline)
ali@0	603	printf(" --> Duplicate footer?\n");
ali@40	604	}
ali@40	605	else
ali@41	606	results.footerline=linecnt;
ali@40	607	}
ali@40	608	}
ali@40	609	}
ali@40	610	if (spline)
ali@41	611	results.firstline=spline;
ali@40	612	if (nspline)
ali@41	613	results.firstline=nspline; /* override with new */
ali@41	614	if (results.footerline)
ali@40	615	continue; /* don't count the boilerplate in the footer */
ali@40	616	llen=strlen(aline);
ali@41	617	results.totlen+=llen;
ali@40	618	for (i=0;i<llen;i++)
ali@40	619	{
ali@40	620	if ((unsigned char)aline[i]>127)
ali@41	621	results.binlen++;
ali@40	622	if (gcisalpha(aline[i]))
ali@41	623	results.alphalen++;
ali@40	624	if (i>0 && aline[i]==CHAR_DQUOTE && isalpha(aline[i-1]))
ali@41	625	results.endquote_count++;
ali@40	626	}
ali@40	627	if (strlen(aline)>2 && lastlen>2 && lastlen<SHORTEST_PG_LINE &&
ali@40	628	lastblen>2 && lastblen>SHORTEST_PG_LINE && laststart!=CHAR_SPACE)
ali@41	629	results.shortline++;
ali@40	630	if (*aline && (unsigned char)aline[strlen(aline)-1]<=CHAR_SPACE)
ali@40	631	cnt_spacend++;
ali@40	632	if (strstr(aline,".,"))
ali@41	633	results.dotcomma++;
ali@40	634	/* only count ast lines for ignoring purposes where there is */
ali@0	635	/* locase text on the line */
ali@40	636	if (strstr(aline,"*"))
ali@40	637	{
ali@40	638	for (s=aline;*s;s++)
ali@40	639	if (s>='a' && s<='z')
ali@0	640	break;
ali@40	641	if (*s)
ali@41	642	results.astline++;
ali@40	643	}
ali@40	644	if (strstr(aline,"/"))
ali@41	645	results.fslashline++;
ali@40	646	for (i=llen-1;i>0 && (unsigned char)aline[i]<=CHAR_SPACE;i--)
ali@40	647	;
ali@40	648	if (aline[i]=='-' && aline[i-1]!='-')
ali@41	649	results.hyphens++;
ali@40	650	if (llen>LONGEST_PG_LINE)
ali@41	651	results.longline++;
ali@40	652	if (llen>WAY_TOO_LONG)
ali@41	653	results.verylongline++;
ali@40	654	if (strstr(aline,"<") && strstr(aline,">"))
ali@40	655	{
ali@40	656	i=(signed int)(strstr(aline,">")-strstr(aline,"<")+1);
ali@40	657	if (i>0)
ali@41	658	results.htmcount++;
ali@40	659	if (strstr(aline,"<i>"))
ali@41	660	results.htmcount+=4; /* bonus marks! */
ali@40	661	}
ali@0	662	/* Check for spaced em-dashes */
ali@40	663	if (strstr(aline,"--"))
ali@40	664	{
ali@41	665	results.emdash++;
ali@40	666	if (*(strstr(aline,"--")-1)==CHAR_SPACE \|\|
ali@40	667	(*(strstr(aline,"--")+2)==CHAR_SPACE))
ali@41	668	results.space_emdash++;
ali@40	669	if (*(strstr(aline,"--")-1)==CHAR_SPACE &&
ali@40	670	(*(strstr(aline,"--")+2)==CHAR_SPACE))
ali@40	671	/* count of em-dashes with spaces both sides */
ali@41	672	results.non_PG_space_emdash++;
ali@40	673	if (*(strstr(aline,"--")-1)!=CHAR_SPACE &&
ali@40	674	(*(strstr(aline,"--")+2)!=CHAR_SPACE))
ali@40	675	/* count of PG-type em-dashes with no spaces */
ali@41	676	results.PG_space_emdash++;
ali@40	677	}
ali@40	678	for (s=aline;*s;)
ali@40	679	{
ali@40	680	s=getaword(s,inword);
ali@40	681	if (!strcmp(inword,"hij") \|\| !strcmp(inword,"niet"))
ali@41	682	results.Dutchcount++;
ali@40	683	if (!strcmp(inword,"dans") \|\| !strcmp(inword,"avec"))
ali@41	684	results.Frenchcount++;
ali@40	685	if (!strcmp(inword,"0") \|\| !strcmp(inword,"1"))
ali@41	686	results.standalone_digit++;
ali@40	687	}
ali@0	688	/* Check for spaced dashes */
ali@40	689	if (strstr(aline," -") && *(strstr(aline," -")+2)!='-')
ali@41	690	results.spacedash++;
ali@40	691	lastblen=lastlen;
ali@40	692	lastlen=strlen(aline);
ali@40	693	laststart=aline[0];
ali@40	694	}
ali@41	695	return &results;
ali@41	696	}
ali@41	697
ali@42	698	struct warnings {
ali@42	699	signed int shortline,longline,bin,dash,dotcomma,ast,fslash,digit,hyphen;
ali@42	700	signed int endquote,isDutch,isFrench;
ali@42	701	};
ali@42	702
ali@42	703	/*
ali@42	704	* report_first_pass:
ali@42	705	*
ali@42	706	* Make some snap decisions based on the first pass results.
ali@42	707	*/
ali@42	708	struct warnings report_first_pass(struct first_pass_results results)
ali@42	709	{
ali@42	710	static struct warnings warnings={0};
ali@42	711	if (cnt_spacend>0)
ali@42	712	printf(" --> %ld lines in this file have white space at end\n",
ali@42	713	cnt_spacend);
ali@42	714	warnings.dotcomma=1;
ali@42	715	if (results->dotcomma>5)
ali@42	716	{
ali@42	717	warnings.dotcomma=0;
ali@42	718	printf(" --> %ld lines in this file contain '.,'. "
ali@42	719	"Not reporting them.\n",results->dotcomma);
ali@42	720	}
ali@42	721	/*
ali@42	722	* If more than 50 lines, or one-tenth, are short,
ali@42	723	* don't bother reporting them.
ali@42	724	*/
ali@42	725	warnings.shortline=1;
ali@42	726	if (results->shortline>50 \|\| results->shortline*10>linecnt)
ali@42	727	{
ali@42	728	warnings.shortline=0;
ali@42	729	printf(" --> %ld lines in this file are short. "
ali@42	730	"Not reporting short lines.\n",results->shortline);
ali@42	731	}
ali@42	732	/*
ali@42	733	* If more than 50 lines, or one-tenth, are long,
ali@42	734	* don't bother reporting them.
ali@42	735	*/
ali@42	736	warnings.longline=1;
ali@42	737	if (results->longline>50 \|\| results->longline*10>linecnt)
ali@42	738	{
ali@42	739	warnings.longline=0;
ali@42	740	printf(" --> %ld lines in this file are long. "
ali@42	741	"Not reporting long lines.\n",results->longline);
ali@42	742	}
ali@42	743	/* If more than 10 lines contain asterisks, don't bother reporting them. */
ali@42	744	warnings.ast=1;
ali@42	745	if (results->astline>10)
ali@42	746	{
ali@42	747	warnings.ast=0;
ali@42	748	printf(" --> %ld lines in this file contain asterisks. "
ali@42	749	"Not reporting them.\n",results->astline);
ali@42	750	}
ali@42	751	/*
ali@42	752	* If more than 10 lines contain forward slashes,
ali@42	753	* don't bother reporting them.
ali@42	754	*/
ali@42	755	warnings.fslash=1;
ali@42	756	if (results->fslashline>10)
ali@42	757	{
ali@42	758	warnings.fslash=0;
ali@42	759	printf(" --> %ld lines in this file contain forward slashes. "
ali@42	760	"Not reporting them.\n",results->fslashline);
ali@42	761	}
ali@42	762	/*
ali@42	763	* If more than 20 lines contain unpunctuated endquotes,
ali@42	764	* don't bother reporting them.
ali@42	765	*/
ali@42	766	warnings.endquote=1;
ali@42	767	if (results->endquote_count>20)
ali@42	768	{
ali@42	769	warnings.endquote=0;
ali@42	770	printf(" --> %ld lines in this file contain unpunctuated endquotes. "
ali@42	771	"Not reporting them.\n",results->endquote_count);
ali@42	772	}
ali@42	773	/*
ali@42	774	* If more than 15 lines contain standalone digits,
ali@42	775	* don't bother reporting them.
ali@42	776	*/
ali@42	777	warnings.digit=1;
ali@42	778	if (results->standalone_digit>10)
ali@42	779	{
ali@42	780	warnings.digit=0;
ali@42	781	printf(" --> %ld lines in this file contain standalone 0s and 1s. "
ali@42	782	"Not reporting them.\n",results->standalone_digit);
ali@42	783	}
ali@42	784	/*
ali@42	785	* If more than 20 lines contain hyphens at end,
ali@42	786	* don't bother reporting them.
ali@42	787	*/
ali@42	788	warnings.hyphen=1;
ali@42	789	if (results->hyphens>20)
ali@42	790	{
ali@42	791	warnings.hyphen=0;
ali@42	792	printf(" --> %ld lines in this file have hyphens at end. "
ali@42	793	"Not reporting them.\n",results->hyphens);
ali@42	794	}
ali@42	795	if (results->htmcount>20 && !pswit[MARKUP_SWITCH])
ali@42	796	{
ali@42	797	printf(" --> Looks like this is HTML. Switching HTML mode ON.\n");
ali@42	798	pswit[MARKUP_SWITCH]=1;
ali@42	799	}
ali@42	800	if (results->verylongline>0)
ali@42	801	printf(" --> %ld lines in this file are VERY long!\n",
ali@42	802	results->verylongline);
ali@42	803	/*
ali@42	804	* If there are more non-PG spaced dashes than PG em-dashes,
ali@42	805	* assume it's deliberate.
ali@42	806	* Current PG guidelines say don't use them, but older texts do,
ali@42	807	* and some people insist on them whatever the guidelines say.
ali@42	808	*/
ali@42	809	warnings.dash=1;
ali@42	810	if (results->spacedash+results->non_PG_space_emdash>
ali@42	811	results->PG_space_emdash)
ali@42	812	{
ali@42	813	warnings.dash=0;
ali@42	814	printf(" --> There are %ld spaced dashes and em-dashes. "
ali@42	815	"Not reporting them.\n",
ali@42	816	results->spacedash+results->non_PG_space_emdash);
ali@42	817	}
ali@42	818	/* If more than a quarter of characters are hi-bit, bug out. */
ali@42	819	warnings.bin=1;
ali@42	820	if (results->binlen*4>results->totlen)
ali@42	821	{
ali@42	822	printf(" --> This file does not appear to be ASCII. "
ali@42	823	"Terminating. Best of luck with it!\n");
ali@42	824	exit(1);
ali@42	825	}
ali@42	826	if (results->alphalen*4<results->totlen)
ali@42	827	{
ali@42	828	printf(" --> This file does not appear to be text. "
ali@42	829	"Terminating. Best of luck with it!\n");
ali@42	830	exit(1);
ali@42	831	}
ali@42	832	if (results->binlen*100>results->totlen \|\| results->binlen>100)
ali@42	833	{
ali@42	834	printf(" --> There are a lot of foreign letters here. "
ali@42	835	"Not reporting them.\n");
ali@42	836	warnings.bin=0;
ali@42	837	}
ali@42	838	warnings.isDutch=0;
ali@42	839	if (results->Dutchcount>50)
ali@42	840	{
ali@42	841	warnings.isDutch=1;
ali@42	842	printf(" --> This looks like Dutch - "
ali@42	843	"switching off dashes and warnings for 's Middags case.\n");
ali@42	844	}
ali@42	845	warnings.isFrench=0;
ali@42	846	if (results->Frenchcount>50)
ali@42	847	{
ali@42	848	warnings.isFrench=1;
ali@42	849	printf(" --> This looks like French - "
ali@42	850	"switching off some doublepunct.\n");
ali@42	851	}
ali@42	852	if (results->firstline && results->footerline)
ali@42	853	printf(" The PG header and footer appear to be already on.\n");
ali@42	854	else
ali@42	855	{
ali@42	856	if (results->firstline)
ali@42	857	printf(" The PG header is on - no footer.\n");
ali@42	858	if (results->footerline)
ali@42	859	printf(" The PG footer is on - no header.\n");
ali@42	860	}
ali@42	861	printf("\n");
ali@42	862	if (pswit[VERBOSE_SWITCH])
ali@42	863	{
ali@42	864	warnings.bin=1;
ali@42	865	warnings.shortline=1;
ali@42	866	warnings.dotcomma=1;
ali@42	867	warnings.longline=1;
ali@42	868	warnings.dash=1;
ali@42	869	warnings.digit=1;
ali@42	870	warnings.ast=1;
ali@42	871	warnings.fslash=1;
ali@42	872	warnings.hyphen=1;
ali@42	873	warnings.endquote=1;
ali@42	874	printf(" * Verbose output is ON -- you asked for it! *\n");
ali@42	875	}
ali@42	876	if (warnings.isDutch)
ali@42	877	warnings.dash=0;
ali@42	878	if (results->footerline>0 && results->firstline>0 &&
ali@42	879	results->footerline>results->firstline &&
ali@42	880	results->footerline-results->firstline<100)
ali@42	881	{
ali@42	882	printf(" --> I don't really know where this text starts. \n");
ali@42	883	printf(" There are no reference points.\n");
ali@42	884	printf(" I'm going to have to report the header and footer "
ali@42	885	"as well.\n");
ali@42	886	results->firstline=0;
ali@42	887	}
ali@42	888	return &warnings;
ali@42	889	}
ali@42	890
ali@43	891	struct counters {
ali@43	892	long quot;
ali@43	893	signed int c_unders,c_brack,s_brack,r_brack;
ali@43	894	signed int open_single_quote,close_single_quote;
ali@43	895	};
ali@43	896
ali@43	897	/*
ali@43	898	* analyse_quotes:
ali@43	899	*
ali@43	900	* Look along the line, accumulate the count of quotes, and see
ali@43	901	* if this is an empty line - i.e. a line with nothing on it
ali@43	902	* but spaces.
ali@43	903	* If line has just spaces, period, * and/or - on it, don't
ali@43	904	* count it, since empty lines with asterisks or dashes to
ali@43	905	* separate sections are common.
ali@43	906	*
ali@43	907	* Returns: Non-zero if the line is empty.
ali@43	908	*/
ali@43	909	int analyse_quotes(const char s,struct counters counters)
ali@43	910	{
ali@43	911	signed int guessquote=0;
ali@43	912	int isemptyline=1; /* assume the line is empty until proven otherwise */
ali@43	913	while (*s)
ali@43	914	{
ali@43	915	if (*s==CHAR_DQUOTE)
ali@43	916	counters->quot++;
ali@43	917	if (s==CHAR_SQUOTE \|\| s==CHAR_OPEN_SQUOTE)
ali@43	918	{
ali@43	919	if (s==aline)
ali@43	920	{
ali@43	921	/*
ali@43	922	* At start of line, it can only be an openquote.
ali@43	923	* Hardcode a very common exception!
ali@43	924	*/
ali@43	925	if (strncmp(s+2,"tis",3) && strncmp(s+2,"Tis",3))
ali@43	926	counters->open_single_quote++;
ali@43	927	}
ali@43	928	else if (gcisalpha(s[-1]) && gcisalpha(s[1]))
ali@43	929	/* Do nothing! it's definitely an apostrophe, not a quote */
ali@43	930	;
ali@43	931	/* it's outside a word - let's check it out */
ali@43	932	else if (*s==CHAR_OPEN_SQUOTE \|\| gcisalpha(s[1]))
ali@43	933	{
ali@43	934	/* it damwell better BE an openquote */
ali@43	935	if (strncmp(s+1,"tis",3) && strncmp(s+1,"Tis",3))
ali@43	936	/* hardcode a very common exception! */
ali@43	937	counters->open_single_quote++;
ali@43	938	}
ali@43	939	else
ali@43	940	{
ali@43	941	/* now - is it a closequote? */
ali@43	942	guessquote=0; /* accumulate clues */
ali@43	943	if (gcisalpha(s[-1]))
ali@43	944	{
ali@43	945	/* it follows a letter - could be either */
ali@43	946	guessquote++;
ali@43	947	if (s[-1]=='s')
ali@43	948	{
ali@43	949	/* looks like a plural apostrophe */
ali@43	950	guessquote-=3;
ali@43	951	if (s[1]==CHAR_SPACE) /* bonus marks! */
ali@43	952	guessquote-=2;
ali@43	953	}
ali@43	954	}
ali@43	955	/* it doesn't have a letter either side */
ali@43	956	else if (strchr(".?!,;:",s[-1]) && strchr(".?!,;: ",s[1]))
ali@43	957	guessquote+=8; /* looks like a closequote */
ali@43	958	else
ali@43	959	guessquote++;
ali@43	960	if (counters->open_single_quote>counters->close_single_quote)
ali@43	961	/*
ali@43	962	* Give it the benefit of some doubt,
ali@43	963	* if a squote is already open.
ali@43	964	*/
ali@43	965	guessquote++;
ali@43	966	else
ali@43	967	guessquote--;
ali@43	968	if (guessquote>=0)
ali@43	969	counters->close_single_quote++;
ali@43	970	}
ali@43	971	}
ali@43	972	if (s!=CHAR_SPACE && s!='-' && s!='.' && s!=CHAR_ASTERISK &&
ali@43	973	s!=13 && s!=10)
ali@43	974	isemptyline=0; /* ignore lines like * * * as spacers */
ali@43	975	if (*s==CHAR_UNDERSCORE)
ali@43	976	counters->c_unders++;
ali@43	977	if (*s==CHAR_OPEN_CBRACK)
ali@43	978	counters->c_brack++;
ali@43	979	if (*s==CHAR_CLOSE_CBRACK)
ali@43	980	counters->c_brack--;
ali@43	981	if (*s==CHAR_OPEN_RBRACK)
ali@43	982	counters->r_brack++;
ali@43	983	if (*s==CHAR_CLOSE_RBRACK)
ali@43	984	counters->r_brack--;
ali@43	985	if (*s==CHAR_OPEN_SBRACK)
ali@43	986	counters->s_brack++;
ali@43	987	if (*s==CHAR_CLOSE_SBRACK)
ali@43	988	counters->s_brack--;
ali@43	989	s++;
ali@43	990	}
ali@43	991	return isemptyline;
ali@43	992	}
ali@43	993
ali@41	994	/*
ali@67	995	* check_for_control_characters:
ali@67	996	*
ali@67	997	* Check for invalid or questionable characters in the line
ali@67	998	* Anything above 127 is invalid for plain ASCII, and
ali@67	999	* non-printable control characters should also be flagged.
ali@67	1000	* Tabs should generally not be there.
ali@67	1001	*/
ali@67	1002	void check_for_control_characters(const char *aline)
ali@67	1003	{
ali@67	1004	unsigned char c;
ali@67	1005	const char *s;
ali@67	1006	for (s=aline;*s;s++)
ali@67	1007	{
ali@67	1008	c=(unsigned char )s;
ali@67	1009	if (c<CHAR_SPACE && c!=CHAR_LF && c!=CHAR_CR && c!=CHAR_TAB)
ali@67	1010	{
ali@67	1011	if (pswit[ECHO_SWITCH])
ali@67	1012	printf("\n%s\n",aline);
ali@67	1013	if (!pswit[OVERVIEW_SWITCH])
ali@67	1014	printf(" Line %ld column %d - Control character %d\n",
ali@67	1015	linecnt,(int)(s-aline)+1,c);
ali@67	1016	else
ali@67	1017	cnt_bin++;
ali@67	1018	}
ali@67	1019	}
ali@67	1020	}
ali@67	1021
ali@67	1022	/*
ali@44	1023	* check_for_odd_characters:
ali@44	1024	*
ali@44	1025	* Check for binary and other odd characters.
ali@44	1026	*/
ali@44	1027	void check_for_odd_characters(const char aline,const struct warnings warnings,
ali@44	1028	int isemptyline)
ali@44	1029	{
ali@44	1030	/* Don't repeat multiple warnings on one line. */
ali@44	1031	signed int eNon_A=0,eTab=0,eTilde=0,eCarat=0,eFSlash=0,eAst=0;
ali@44	1032	const char *s;
ali@44	1033	unsigned char c;
ali@44	1034	for (s=aline;*s;s++)
ali@44	1035	{
ali@44	1036	c=(unsigned char )s;
ali@44	1037	if (!eNon_A && (s<CHAR_SPACE && s!=9 && *s!='\n' \|\| c>127))
ali@44	1038	{
ali@44	1039	if (pswit[ECHO_SWITCH])
ali@44	1040	printf("\n%s\n",aline);
ali@44	1041	if (!pswit[OVERVIEW_SWITCH])
ali@44	1042	if (c>127 && c<160)
ali@44	1043	printf(" Line %ld column %d - "
ali@44	1044	"Non-ISO-8859 character %d\n",linecnt,(int)(s-aline)+1,c);
ali@44	1045	else
ali@44	1046	printf(" Line %ld column %d - Non-ASCII character %d\n",
ali@44	1047	linecnt,(int)(s-aline)+1,c);
ali@44	1048	else
ali@44	1049	cnt_bin++;
ali@44	1050	eNon_A=1;
ali@44	1051	}
ali@44	1052	if (!eTab && *s==CHAR_TAB)
ali@44	1053	{
ali@44	1054	if (pswit[ECHO_SWITCH])
ali@44	1055	printf("\n%s\n",aline);
ali@44	1056	if (!pswit[OVERVIEW_SWITCH])
ali@44	1057	printf(" Line %ld column %d - Tab character?\n",
ali@44	1058	linecnt,(int)(s-aline)+1);
ali@44	1059	else
ali@44	1060	cnt_odd++;
ali@44	1061	eTab=1;
ali@44	1062	}
ali@44	1063	if (!eTilde && *s==CHAR_TILDE)
ali@44	1064	{
ali@44	1065	/*
ali@44	1066	* Often used by OCR software to indicate an
ali@44	1067	* unrecognizable character.
ali@44	1068	*/
ali@44	1069	if (pswit[ECHO_SWITCH])
ali@44	1070	printf("\n%s\n",aline);
ali@44	1071	if (!pswit[OVERVIEW_SWITCH])
ali@44	1072	printf(" Line %ld column %d - Tilde character?\n",
ali@44	1073	linecnt,(int)(s-aline)+1);
ali@44	1074	else
ali@44	1075	cnt_odd++;
ali@44	1076	eTilde=1;
ali@44	1077	}
ali@44	1078	if (!eCarat && *s==CHAR_CARAT)
ali@44	1079	{
ali@44	1080	if (pswit[ECHO_SWITCH])
ali@44	1081	printf("\n%s\n",aline);
ali@44	1082	if (!pswit[OVERVIEW_SWITCH])
ali@44	1083	printf(" Line %ld column %d - Carat character?\n",
ali@44	1084	linecnt,(int)(s-aline)+1);
ali@44	1085	else
ali@44	1086	cnt_odd++;
ali@44	1087	eCarat=1;
ali@44	1088	}
ali@44	1089	if (!eFSlash && *s==CHAR_FORESLASH && warnings->fslash)
ali@44	1090	{
ali@44	1091	if (pswit[ECHO_SWITCH])
ali@44	1092	printf("\n%s\n",aline);
ali@44	1093	if (!pswit[OVERVIEW_SWITCH])
ali@44	1094	printf(" Line %ld column %d - Forward slash?\n",
ali@44	1095	linecnt,(int)(s-aline)+1);
ali@44	1096	else
ali@44	1097	cnt_odd++;
ali@44	1098	eFSlash=1;
ali@44	1099	}
ali@44	1100	/*
ali@44	1101	* Report asterisks only in paranoid mode,
ali@44	1102	* since they're often deliberate.
ali@44	1103	*/
ali@44	1104	if (!eAst && pswit[PARANOID_SWITCH] && warnings->ast && !isemptyline &&
ali@44	1105	*s==CHAR_ASTERISK)
ali@44	1106	{
ali@44	1107	if (pswit[ECHO_SWITCH])
ali@44	1108	printf("\n%s\n",aline);
ali@44	1109	if (!pswit[OVERVIEW_SWITCH])
ali@44	1110	printf(" Line %ld column %d - Asterisk?\n",
ali@44	1111	linecnt,(int)(s-aline)+1);
ali@44	1112	else
ali@44	1113	cnt_odd++;
ali@44	1114	eAst=1;
ali@44	1115	}
ali@44	1116	}
ali@44	1117	}
ali@44	1118
ali@44	1119	/*
ali@45	1120	* check_for_long_line:
ali@45	1121	*
ali@45	1122	* Check for line too long.
ali@45	1123	*/
ali@45	1124	void check_for_long_line(const char *aline)
ali@45	1125	{
ali@45	1126	if (strlen(aline)>LONGEST_PG_LINE)
ali@45	1127	{
ali@45	1128	if (pswit[ECHO_SWITCH])
ali@45	1129	printf("\n%s\n",aline);
ali@45	1130	if (!pswit[OVERVIEW_SWITCH])
ali@45	1131	printf(" Line %ld column %d - Long line %d\n",
ali@45	1132	linecnt,strlen(aline),strlen(aline));
ali@45	1133	else
ali@45	1134	cnt_long++;
ali@45	1135	}
ali@45	1136	}
ali@45	1137
ali@45	1138	struct line_properties {
ali@45	1139	unsigned int len,blen;
ali@45	1140	char start;
ali@45	1141	};
ali@45	1142
ali@45	1143	/*
ali@45	1144	* check_for_short_line:
ali@45	1145	*
ali@45	1146	* Check for line too short.
ali@45	1147	*
ali@45	1148	* This one is a bit trickier to implement: we don't want to
ali@45	1149	* flag the last line of a paragraph for being short, so we
ali@45	1150	* have to wait until we know that our current line is a
ali@45	1151	* "normal" line, then report the _previous_ line if it was too
ali@45	1152	* short. We also don't want to report indented lines like
ali@45	1153	* chapter heads or formatted quotations. We therefore keep
ali@45	1154	* last->len as the length of the last line examined, and
ali@45	1155	* last->blen as the length of the last but one, and try to
ali@45	1156	* suppress unnecessary warnings by checking that both were of
ali@45	1157	* "normal" length. We keep the first character of the last
ali@45	1158	* line in last->start, and if it was a space, we assume that
ali@45	1159	* the formatting is deliberate. I can't figure out a way to
ali@45	1160	* distinguish something like a quoted verse left-aligned or
ali@45	1161	* the header or footer of a letter from a paragraph of short
ali@45	1162	* lines - maybe if I examined the whole paragraph, and if the
ali@45	1163	* para has less than, say, 8 lines and if all lines are short,
ali@45	1164	* then just assume it's OK? Need to look at some texts to see
ali@45	1165	* how often a formula like this would get the right result.
ali@45	1166	*/
ali@45	1167	void check_for_short_line(const char aline,const struct line_properties last)
ali@45	1168	{
ali@45	1169	if (strlen(aline)>1 && last->len>1 && last->len<SHORTEST_PG_LINE &&
ali@45	1170	last->blen>1 && last->blen>SHORTEST_PG_LINE && last->start!=CHAR_SPACE)
ali@45	1171	{
ali@45	1172	if (pswit[ECHO_SWITCH])
ali@45	1173	printf("\n%s\n",prevline);
ali@45	1174	if (!pswit[OVERVIEW_SWITCH])
ali@45	1175	printf(" Line %ld column %d - Short line %d?\n",
ali@45	1176	linecnt-1,strlen(prevline),strlen(prevline));
ali@45	1177	else
ali@45	1178	cnt_short++;
ali@45	1179	}
ali@45	1180	}
ali@45	1181
ali@45	1182	/*
ali@46	1183	* check_for_starting_punctuation:
ali@46	1184	*
ali@46	1185	* Look for punctuation other than full ellipses at start of line.
ali@46	1186	*/
ali@46	1187	void check_for_starting_punctuation(const char *aline)
ali@46	1188	{
ali@46	1189	if (*aline && strchr(".?!,;:",aline[0]) && strncmp(". . .",aline,5))
ali@46	1190	{
ali@46	1191	if (pswit[ECHO_SWITCH])
ali@46	1192	printf("\n%s\n",aline);
ali@46	1193	if (!pswit[OVERVIEW_SWITCH])
ali@46	1194	printf(" Line %ld column 1 - Begins with punctuation?\n",
ali@46	1195	linecnt);
ali@46	1196	else
ali@46	1197	cnt_punct++;
ali@46	1198	}
ali@46	1199	}
ali@46	1200
ali@46	1201	/*
ali@47	1202	* check_for_spaced_emdash:
ali@47	1203	*
ali@47	1204	* Check for spaced em-dashes.
ali@47	1205	*
ali@47	1206	* We must check _all_ occurrences of "--" on the line
ali@47	1207	* hence the loop - even if the first double-dash is OK
ali@47	1208	* there may be another that's wrong later on.
ali@47	1209	*/
ali@47	1210	void check_for_spaced_emdash(const char *aline)
ali@47	1211	{
ali@47	1212	const char s,t;
ali@47	1213	s=aline;
ali@47	1214	while ((t=strstr(s,"--")))
ali@47	1215	{
ali@47	1216	if (t>aline && t[-1]==CHAR_SPACE \|\| t[2]==CHAR_SPACE)
ali@47	1217	{
ali@47	1218	if (pswit[ECHO_SWITCH])
ali@47	1219	printf("\n%s\n",aline);
ali@47	1220	if (!pswit[OVERVIEW_SWITCH])
ali@47	1221	printf(" Line %ld column %d - Spaced em-dash?\n",
ali@47	1222	linecnt,(int)(t-aline)+1);
ali@47	1223	else
ali@47	1224	cnt_dash++;
ali@47	1225	}
ali@47	1226	s=t+2;
ali@47	1227	}
ali@47	1228	}
ali@47	1229
ali@47	1230	/*
ali@47	1231	* check_for_spaced_dash:
ali@47	1232	*
ali@47	1233	* Check for spaced dashes.
ali@47	1234	*/
ali@47	1235	void check_for_spaced_dash(const char *aline)
ali@47	1236	{
ali@47	1237	const char *s;
ali@47	1238	if ((s=strstr(aline," -")))
ali@47	1239	{
ali@47	1240	if (s[2]!='-')
ali@47	1241	{
ali@47	1242	if (pswit[ECHO_SWITCH])
ali@47	1243	printf("\n%s\n",aline);
ali@47	1244	if (!pswit[OVERVIEW_SWITCH])
ali@47	1245	printf(" Line %ld column %d - Spaced dash?\n",
ali@47	1246	linecnt,(int)(s-aline)+1);
ali@47	1247	else
ali@47	1248	cnt_dash++;
ali@47	1249	}
ali@47	1250	}
ali@47	1251	else if ((s=strstr(aline,"- ")))
ali@47	1252	{
ali@47	1253	if (s==aline \|\| s[-1]!='-')
ali@47	1254	{
ali@47	1255	if (pswit[ECHO_SWITCH])
ali@47	1256	printf("\n%s\n",aline);
ali@47	1257	if (!pswit[OVERVIEW_SWITCH])
ali@47	1258	printf(" Line %ld column %d - Spaced dash?\n",
ali@47	1259	linecnt,(int)(s-aline)+1);
ali@47	1260	else
ali@47	1261	cnt_dash++;
ali@47	1262	}
ali@47	1263	}
ali@47	1264	}
ali@47	1265
ali@47	1266	/*
ali@48	1267	* check_for_unmarked_paragraphs:
ali@48	1268	*
ali@48	1269	* Check for unmarked paragraphs indicated by separate speakers.
ali@48	1270	*
ali@48	1271	* May well be false positive:
ali@48	1272	* "Bravo!" "Wonderful!" called the crowd.
ali@48	1273	* but useful all the same.
ali@48	1274	*/
ali@48	1275	void check_for_unmarked_paragraphs(const char *aline)
ali@48	1276	{
ali@48	1277	const char *s;
ali@48	1278	s=strstr(aline,"\" \"");
ali@48	1279	if (!s)
ali@48	1280	s=strstr(aline,"\" \"");
ali@48	1281	if (s)
ali@48	1282	{
ali@48	1283	if (pswit[ECHO_SWITCH])
ali@48	1284	printf("\n%s\n",aline);
ali@48	1285	if (!pswit[OVERVIEW_SWITCH])
ali@48	1286	printf(" Line %ld column %d - Query missing paragraph break?\n",
ali@48	1287	linecnt,(int)(s-aline)+1);
ali@48	1288	else
ali@48	1289	cnt_punct++;
ali@48	1290	}
ali@48	1291	}
ali@48	1292
ali@48	1293	/*
ali@49	1294	* check_for_jeebies:
ali@49	1295	*
ali@49	1296	* Check for "to he" and other easy h/b errors.
ali@49	1297	*
ali@49	1298	* This is a very inadequate effort on the h/b problem,
ali@49	1299	* but the phrase "to he" is always an error, whereas "to
ali@49	1300	* be" is quite common.
ali@49	1301	* Similarly, '"Quiet!", be said.' is a non-be error
ali@49	1302	* "to he" is _not_ always an error!:
ali@49	1303	* "Where they went to he couldn't say."
ali@49	1304	* Another false positive:
ali@49	1305	* What would "Cinderella" be without the . . .
ali@49	1306	* and another: "If he wants to he can see for himself."
ali@49	1307	*/
ali@49	1308	void check_for_jeebies(const char *aline)
ali@49	1309	{
ali@49	1310	const char *s;
ali@49	1311	s=strstr(aline," be could ");
ali@49	1312	if (!s)
ali@49	1313	s=strstr(aline," be would ");
ali@49	1314	if (!s)
ali@49	1315	s=strstr(aline," was be ");
ali@49	1316	if (!s)
ali@49	1317	s=strstr(aline," be is ");
ali@49	1318	if (!s)
ali@49	1319	s=strstr(aline," is be ");
ali@49	1320	if (!s)
ali@49	1321	s=strstr(aline,"\", be ");
ali@49	1322	if (!s)
ali@49	1323	s=strstr(aline,"\" be ");
ali@49	1324	if (!s)
ali@49	1325	s=strstr(aline,"\" be ");
ali@49	1326	if (!s)
ali@49	1327	s=strstr(aline," to he ");
ali@49	1328	if (s)
ali@49	1329	{
ali@49	1330	if (pswit[ECHO_SWITCH])
ali@49	1331	printf("\n%s\n",aline);
ali@49	1332	if (!pswit[OVERVIEW_SWITCH])
ali@49	1333	printf(" Line %ld column %d - Query he/be error?\n",
ali@49	1334	linecnt,(int)(s-aline)+1);
ali@49	1335	else
ali@49	1336	cnt_word++;
ali@49	1337	}
ali@49	1338	s=strstr(aline," the had ");
ali@49	1339	if (!s)
ali@49	1340	s=strstr(aline," a had ");
ali@49	1341	if (!s)
ali@49	1342	s=strstr(aline," they bad ");
ali@49	1343	if (!s)
ali@49	1344	s=strstr(aline," she bad ");
ali@49	1345	if (!s)
ali@49	1346	s=strstr(aline," he bad ");
ali@49	1347	if (!s)
ali@49	1348	s=strstr(aline," you bad ");
ali@49	1349	if (!s)
ali@49	1350	s=strstr(aline," i bad ");
ali@49	1351	if (s)
ali@49	1352	{
ali@49	1353	if (pswit[ECHO_SWITCH])
ali@49	1354	printf("\n%s\n",aline);
ali@49	1355	if (!pswit[OVERVIEW_SWITCH])
ali@49	1356	printf(" Line %ld column %d - Query had/bad error?\n",
ali@49	1357	linecnt,(int)(s-aline)+1);
ali@49	1358	else
ali@49	1359	cnt_word++;
ali@49	1360	}
ali@49	1361	s=strstr(aline,"; hut ");
ali@49	1362	if (!s)
ali@49	1363	s=strstr(aline,", hut ");
ali@49	1364	if (s)
ali@49	1365	{
ali@49	1366	if (pswit[ECHO_SWITCH])
ali@49	1367	printf("\n%s\n",aline);
ali@49	1368	if (!pswit[OVERVIEW_SWITCH])
ali@49	1369	printf(" Line %ld column %d - Query hut/but error?\n",
ali@49	1370	linecnt,(int)(s-aline)+1);
ali@49	1371	else
ali@49	1372	cnt_word++;
ali@49	1373	}
ali@49	1374	}
ali@49	1375
ali@49	1376	/*
ali@50	1377	* check_for_mta_from:
ali@50	1378	*
ali@50	1379	* Special case - angled bracket in front of "From" placed there by an
ali@50	1380	* MTA when sending an e-mail.
ali@50	1381	*/
ali@50	1382	void check_for_mta_from(const char *aline)
ali@50	1383	{
ali@50	1384	const char *s;
ali@50	1385	s=strstr(aline,">From");
ali@50	1386	if (s)
ali@50	1387	{
ali@50	1388	if (pswit[ECHO_SWITCH])
ali@50	1389	printf("\n%s\n",aline);
ali@50	1390	if (!pswit[OVERVIEW_SWITCH])
ali@50	1391	printf(" Line %ld column %d - Query angled bracket with From\n",
ali@50	1392	linecnt,(int)(s-aline)+1);
ali@50	1393	else
ali@50	1394	cnt_punct++;
ali@50	1395	}
ali@50	1396	}
ali@50	1397
ali@50	1398	/*
ali@51	1399	* check_for_orphan_character:
ali@51	1400	*
ali@51	1401	* Check for a single character line -
ali@51	1402	* often an overflow from bad wrapping.
ali@51	1403	*/
ali@51	1404	void check_for_orphan_character(const char *aline)
ali@51	1405	{
ali@51	1406	if (*aline && !aline[1])
ali@51	1407	{
ali@51	1408	if (aline=='I' \|\| aline=='V' \|\| aline=='X' \|\| aline=='L' \|\|
ali@51	1409	gcisdigit(*aline))
ali@51	1410	; /* Nothing - ignore numerals alone on a line. */
ali@51	1411	else
ali@51	1412	{
ali@51	1413	if (pswit[ECHO_SWITCH])
ali@51	1414	printf("\n%s\n",aline);
ali@51	1415	if (!pswit[OVERVIEW_SWITCH])
ali@51	1416	printf(" Line %ld column 1 - Query single character line\n",
ali@51	1417	linecnt);
ali@51	1418	else
ali@51	1419	cnt_punct++;
ali@51	1420	}
ali@51	1421	}
ali@51	1422	}
ali@51	1423
ali@51	1424	/*
ali@52	1425	* check_for_pling_scanno:
ali@52	1426	*
ali@52	1427	* Check for I" - often should be !
ali@52	1428	*/
ali@52	1429	void check_for_pling_scanno(const char *aline)
ali@52	1430	{
ali@52	1431	const char *s;
ali@52	1432	s=strstr(aline," I\"");
ali@52	1433	if (s)
ali@52	1434	{
ali@52	1435	if (pswit[ECHO_SWITCH])
ali@52	1436	printf("\n%s\n",aline);
ali@52	1437	if (!pswit[OVERVIEW_SWITCH])
ali@52	1438	printf(" Line %ld column %ld - Query I=exclamation mark?\n",
ali@52	1439	linecnt,s-aline);
ali@52	1440	else
ali@52	1441	cnt_punct++;
ali@52	1442	}
ali@52	1443	}
ali@52	1444
ali@52	1445	/*
ali@53	1446	* check_for_extra_period:
ali@53	1447	*
ali@53	1448	* Check for period without a capital letter. Cut-down from gutspell.
ali@53	1449	* Only works when it happens on a single line.
ali@53	1450	*/
ali@53	1451	void check_for_extra_period(const char aline,const struct warnings warnings)
ali@53	1452	{
ali@53	1453	const char s,t,*s1;
ali@53	1454	signed int i,istypo,isdup;
ali@53	1455	static char qperiod[MAX_QWORD][MAX_QWORD_LENGTH];
ali@53	1456	static int qperiod_index=0;
ali@53	1457	char testword[MAXWORDLEN]="";
ali@53	1458	if (pswit[PARANOID_SWITCH])
ali@53	1459	{
ali@53	1460	for (t=s=aline;strstr(t,". ");)
ali@53	1461	{
ali@53	1462	t=strstr(t,". ");
ali@53	1463	if (t==s)
ali@53	1464	{
ali@53	1465	t++;
ali@53	1466	/* start of line punctuation is handled elsewhere */
ali@53	1467	continue;
ali@53	1468	}
ali@53	1469	if (!gcisalpha(t[-1]))
ali@53	1470	{
ali@53	1471	t++;
ali@53	1472	continue;
ali@53	1473	}
ali@53	1474	if (warnings->isDutch)
ali@53	1475	{
ali@53	1476	/* For Frank & Jeroen -- 's Middags case */
ali@53	1477	if (t[2]==CHAR_SQUOTE && t[3]>='a' && t[3]<='z' &&
ali@53	1478	t[4]==CHAR_SPACE && t[5]>='A' && t[5]<='Z')
ali@53	1479	{
ali@53	1480	t++;
ali@53	1481	continue;
ali@53	1482	}
ali@53	1483	}
ali@53	1484	s1=t+2;
ali@53	1485	while (s1 && !gcisalpha(s1) && !isdigit(*s1))
ali@53	1486	s1++;
ali@53	1487	if (s1>='a' && s1<='z')
ali@53	1488	{
ali@53	1489	/* we have something to investigate */
ali@53	1490	istypo=1;
ali@53	1491	/* so let's go back and find out */
ali@53	1492	for (s1=t-1;s1>=s &&
ali@53	1493	(gcisalpha(s1) \|\| gcisdigit(s1) \|\| *s1==CHAR_SQUOTE &&
ali@53	1494	gcisalpha(s1[1]) && gcisalpha(s1[-1]));s1--)
ali@53	1495	;
ali@53	1496	s1++;
ali@53	1497	for (i=0;s1 && s1!='.';s1++,i++)
ali@53	1498	testword[i]=*s1;
ali@53	1499	testword[i]=0;
ali@53	1500	for (i=0;*abbrev[i];i++)
ali@53	1501	if (!strcmp(testword,abbrev[i]))
ali@53	1502	istypo=0;
ali@53	1503	if (gcisdigit(*testword))
ali@53	1504	istypo=0;
ali@53	1505	if (!testword[1])
ali@53	1506	istypo=0;
ali@53	1507	if (isroman(testword))
ali@53	1508	istypo=0;
ali@53	1509	if (istypo)
ali@53	1510	{
ali@53	1511	istypo=0;
ali@53	1512	for (i=0;testword[i];i++)
ali@53	1513	if (strchr(vowels,testword[i]))
ali@53	1514	istypo=1;
ali@53	1515	}
ali@53	1516	if (istypo)
ali@53	1517	{
ali@53	1518	isdup=0;
ali@53	1519	if (strlen(testword)<MAX_QWORD_LENGTH &&
ali@53	1520	!pswit[VERBOSE_SWITCH])
ali@53	1521	for (i=0;i<qperiod_index;i++)
ali@53	1522	if (!strcmp(testword,qperiod[i]))
ali@53	1523	isdup=1;
ali@53	1524	if (!isdup)
ali@53	1525	{
ali@53	1526	if (qperiod_index<MAX_QWORD &&
ali@53	1527	strlen(testword)<MAX_QWORD_LENGTH)
ali@53	1528	{
ali@53	1529	strcpy(qperiod[qperiod_index],testword);
ali@53	1530	qperiod_index++;
ali@53	1531	}
ali@53	1532	if (pswit[ECHO_SWITCH])
ali@53	1533	printf("\n%s\n",aline);
ali@53	1534	if (!pswit[OVERVIEW_SWITCH])
ali@53	1535	printf(" Line %ld column %d - Extra period?\n",
ali@53	1536	linecnt,(int)(t-aline)+1);
ali@53	1537	else
ali@53	1538	cnt_punct++;
ali@53	1539	}
ali@53	1540	}
ali@53	1541	}
ali@53	1542	t++;
ali@53	1543	}
ali@53	1544	}
ali@53	1545	}
ali@53	1546
ali@53	1547	/*
ali@54	1548	* check_for_following_punctuation:
ali@54	1549	*
ali@54	1550	* Check for words usually not followed by punctuation.
ali@54	1551	*/
ali@54	1552	void check_for_following_punctuation(const char *aline)
ali@54	1553	{
ali@54	1554	int i;
ali@54	1555	const char s,wordstart;
ali@54	1556	char inword[MAXWORDLEN];
ali@54	1557	if (pswit[TYPO_SWITCH])
ali@54	1558	{
ali@54	1559	for (s=aline;*s;)
ali@54	1560	{
ali@54	1561	wordstart=s;
ali@54	1562	s=getaword(s,inword);
ali@54	1563	if (!*inword)
ali@54	1564	continue;
ali@54	1565	lowerit(inword);
ali@54	1566	for (i=0;*nocomma[i];i++)
ali@54	1567	if (!strcmp(inword,nocomma[i]))
ali@54	1568	{
ali@54	1569	if (s==',' \|\| s==';' \|\| *s==':')
ali@54	1570	{
ali@54	1571	if (pswit[ECHO_SWITCH])
ali@54	1572	printf("\n%s\n",aline);
ali@54	1573	if (!pswit[OVERVIEW_SWITCH])
ali@54	1574	printf(" Line %ld column %d - "
ali@54	1575	"Query punctuation after %s?\n",
ali@54	1576	linecnt,(int)(s-aline)+1,inword);
ali@54	1577	else
ali@54	1578	cnt_punct++;
ali@54	1579	}
ali@54	1580	}
ali@54	1581	for (i=0;*noperiod[i];i++)
ali@54	1582	if (!strcmp(inword,noperiod[i]))
ali@54	1583	{
ali@54	1584	if (s=='.' \|\| s=='!')
ali@54	1585	{
ali@54	1586	if (pswit[ECHO_SWITCH])
ali@54	1587	printf("\n%s\n",aline);
ali@54	1588	if (!pswit[OVERVIEW_SWITCH])
ali@54	1589	printf(" Line %ld column %d - "
ali@54	1590	"Query punctuation after %s?\n",
ali@54	1591	linecnt,(int)(s-aline)+1,inword);
ali@54	1592	else
ali@54	1593	cnt_punct++;
ali@54	1594	}
ali@54	1595	}
ali@54	1596	}
ali@54	1597	}
ali@54	1598	}
ali@54	1599
ali@54	1600	/*
ali@55	1601	* check_for_typos:
ali@55	1602	*
ali@55	1603	* Check for commonly mistyped words,
ali@55	1604	* and digits like 0 for O in a word.
ali@55	1605	*/
ali@55	1606	void check_for_typos(const char aline,struct warnings warnings)
ali@55	1607	{
ali@55	1608	const char s,wordstart;
ali@55	1609	char inword[MAXWORDLEN],testword[MAXWORDLEN];
ali@55	1610	int i,istypo,isdup,alower,vowel,consonant;
ali@55	1611	static int qword_index=0;
ali@55	1612	for (s=aline;*s;)
ali@55	1613	{
ali@55	1614	wordstart=s;
ali@55	1615	s=getaword(s,inword);
ali@55	1616	if (!*inword)
ali@55	1617	continue; /* don't bother with empty lines */
ali@55	1618	if (mixdigit(inword))
ali@55	1619	{
ali@55	1620	if (pswit[ECHO_SWITCH])
ali@55	1621	printf("\n%s\n",aline);
ali@55	1622	if (!pswit[OVERVIEW_SWITCH])
ali@55	1623	printf(" Line %ld column %d - Query digit in %s\n",
ali@55	1624	linecnt,(int)(wordstart-aline)+1,inword);
ali@55	1625	else
ali@55	1626	cnt_word++;
ali@55	1627	}
ali@55	1628	/*
ali@55	1629	* Put the word through a series of tests for likely typos and OCR
ali@55	1630	* errors.
ali@55	1631	*/
ali@55	1632	if (pswit[TYPO_SWITCH])
ali@55	1633	{
ali@55	1634	istypo=0;
ali@55	1635	strcpy(testword,inword);
ali@55	1636	alower=0;
ali@55	1637	for (i=0;i<(signed int)strlen(testword);i++)
ali@55	1638	{
ali@55	1639	/* lowercase for testing */
ali@55	1640	if (testword[i]>='a' && testword[i]<='z')
ali@55	1641	alower=1;
ali@55	1642	if (alower && testword[i]>='A' && testword[i]<='Z')
ali@55	1643	{
ali@55	1644	/*
ali@55	1645	* We have an uppercase mid-word. However, there are
ali@55	1646	* common cases:
ali@55	1647	* Mac and Mc like McGill
ali@55	1648	* French contractions like l'Abbe
ali@55	1649	*/
ali@55	1650	if (i==2 && testword[0]=='m' && testword[1]=='c' \|\|
ali@55	1651	i==3 && testword[0]=='m' && testword[1]=='a' &&
ali@55	1652	testword[2]=='c' \|\| i>0 && testword[i-1]==CHAR_SQUOTE)
ali@55	1653	; /* do nothing! */
ali@55	1654	else
ali@55	1655	istypo=1;
ali@55	1656	}
ali@55	1657	testword[i]=(char)tolower(testword[i]);
ali@55	1658	}
ali@55	1659	/*
ali@55	1660	* Check for certain unlikely two-letter combinations at word
ali@55	1661	* start and end.
ali@55	1662	*/
ali@55	1663	if (strlen(testword)>1)
ali@55	1664	{
ali@55	1665	for (i=0;*nostart[i];i++)
ali@55	1666	if (!strncmp(testword,nostart[i],2))
ali@55	1667	istypo=1;
ali@55	1668	for (i=0;*noend[i];i++)
ali@55	1669	if (!strncmp(testword+strlen(testword)-2,noend[i],2))
ali@55	1670	istypo=1;
ali@55	1671	}
ali@55	1672	/* ght is common, gbt never. Like that. */
ali@55	1673	if (strstr(testword,"cb"))
ali@55	1674	istypo=1;
ali@55	1675	if (strstr(testword,"gbt"))
ali@55	1676	istypo=1;
ali@55	1677	if (strstr(testword,"pbt"))
ali@55	1678	istypo=1;
ali@55	1679	if (strstr(testword,"tbs"))
ali@55	1680	istypo=1;
ali@55	1681	if (strstr(testword,"mrn"))
ali@55	1682	istypo=1;
ali@55	1683	if (strstr(testword,"ahle"))
ali@55	1684	istypo=1;
ali@55	1685	if (strstr(testword,"ihle"))
ali@55	1686	istypo=1;
ali@55	1687	/*
ali@55	1688	* "TBE" does happen - like HEARTBEAT - but uncommon.
ali@55	1689	* Also "TBI" - frostbite, outbid - but uncommon.
ali@55	1690	* Similarly "ii" like Hawaii, or Pompeii, and in Roman
ali@55	1691	* numerals, but "ii" is a common scanno.
ali@55	1692	*/
ali@55	1693	if (strstr(testword,"tbi"))
ali@55	1694	istypo=1;
ali@55	1695	if (strstr(testword,"tbe"))
ali@55	1696	istypo=1;
ali@55	1697	if (strstr(testword,"ii"))
ali@55	1698	istypo=1;
ali@55	1699	/*
ali@55	1700	* Check for no vowels or no consonants.
ali@55	1701	* If none, flag a typo.
ali@55	1702	*/
ali@55	1703	if (!istypo && strlen(testword)>1)
ali@55	1704	{
ali@55	1705	vowel=consonant=0;
ali@55	1706	for (i=0;testword[i];i++)
ali@55	1707	{
ali@55	1708	if (testword[i]=='y' \|\| gcisdigit(testword[i]))
ali@55	1709	{
ali@55	1710	/* Yah, this is loose. */
ali@55	1711	vowel++;
ali@55	1712	consonant++;
ali@55	1713	}
ali@55	1714	else if (strchr(vowels,testword[i]))
ali@55	1715	vowel++;
ali@55	1716	else
ali@55	1717	consonant++;
ali@55	1718	}
ali@55	1719	if (!vowel \|\| !consonant)
ali@55	1720	istypo=1;
ali@55	1721	}
ali@55	1722	/*
ali@55	1723	* Now exclude the word from being reported if it's in
ali@55	1724	* the okword list.
ali@55	1725	*/
ali@55	1726	for (i=0;*okword[i];i++)
ali@55	1727	if (!strcmp(testword,okword[i]))
ali@55	1728	istypo=0;
ali@55	1729	/*
ali@55	1730	* What looks like a typo may be a Roman numeral.
ali@55	1731	* Exclude these.
ali@55	1732	*/
ali@55	1733	if (istypo && isroman(testword))
ali@55	1734	istypo=0;
ali@55	1735	/* Check the manual list of typos. */
ali@55	1736	if (!istypo)
ali@55	1737	for (i=0;*typo[i];i++)
ali@55	1738	if (!strcmp(testword,typo[i]))
ali@55	1739	istypo=1;
ali@55	1740	/*
ali@55	1741	* Check lowercase s, l, i and m - special cases.
ali@55	1742	* "j" - often a semi-colon gone wrong.
ali@55	1743	* "d" for a missing apostrophe - he d
ali@55	1744	* "n" for "in"
ali@55	1745	*/
ali@55	1746	if (!istypo && strlen(testword)==1 && strchr("slmijdn",*inword))
ali@55	1747	istypo=1;
ali@55	1748	if (istypo)
ali@55	1749	{
ali@55	1750	isdup=0;
ali@55	1751	if (strlen(testword)<MAX_QWORD_LENGTH &&
ali@55	1752	!pswit[VERBOSE_SWITCH])
ali@55	1753	for (i=0;i<qword_index;i++)
ali@55	1754	if (!strcmp(testword,qword[i]))
ali@55	1755	{
ali@55	1756	isdup=1;
ali@55	1757	++dupcnt[i];
ali@55	1758	}
ali@55	1759	if (!isdup)
ali@55	1760	{
ali@55	1761	if (qword_index<MAX_QWORD &&
ali@55	1762	strlen(testword)<MAX_QWORD_LENGTH)
ali@55	1763	{
ali@55	1764	strcpy(qword[qword_index],testword);
ali@55	1765	qword_index++;
ali@55	1766	}
ali@55	1767	if (pswit[ECHO_SWITCH])
ali@55	1768	printf("\n%s\n",aline);
ali@55	1769	if (!pswit[OVERVIEW_SWITCH])
ali@55	1770	{
ali@55	1771	printf(" Line %ld column %d - Query word %s",
ali@55	1772	linecnt,(int)(wordstart-aline)+1,inword);
ali@55	1773	if (strlen(testword)<MAX_QWORD_LENGTH &&
ali@55	1774	!pswit[VERBOSE_SWITCH])
ali@55	1775	printf(" - not reporting duplicates");
ali@55	1776	printf("\n");
ali@55	1777	}
ali@55	1778	else
ali@55	1779	cnt_word++;
ali@55	1780	}
ali@55	1781	}
ali@55	1782	}
ali@55	1783	/* check the user's list of typos */
ali@55	1784	if (!istypo && usertypo_count)
ali@55	1785	for (i=0;i<usertypo_count;i++)
ali@55	1786	if (!strcmp(testword,usertypo[i]))
ali@55	1787	{
ali@55	1788	if (pswit[ECHO_SWITCH])
ali@55	1789	printf("\n%s\n",aline);
ali@55	1790	if (!pswit[OVERVIEW_SWITCH])
ali@55	1791	printf(" Line %ld column %d - "
ali@55	1792	"Query possible scanno %s\n",
ali@55	1793	linecnt,(int)(wordstart-aline)+2,inword);
ali@55	1794	}
ali@55	1795	if (pswit[PARANOID_SWITCH] && warnings->digit)
ali@55	1796	{
ali@55	1797	/* In paranoid mode, query all 0 and 1 standing alone. */
ali@55	1798	if (!strcmp(inword,"0") \|\| !strcmp(inword,"1"))
ali@55	1799	{
ali@55	1800	if (pswit[ECHO_SWITCH])
ali@55	1801	printf("\n%s\n",aline);
ali@55	1802	if (!pswit[OVERVIEW_SWITCH])
ali@55	1803	printf(" Line %ld column %d - Query standalone %s\n",
ali@55	1804	linecnt,(int)(wordstart-aline)+2,inword);
ali@55	1805	else
ali@55	1806	cnt_word++;
ali@55	1807	}
ali@55	1808	}
ali@55	1809	}
ali@55	1810	}
ali@55	1811
ali@56	1812	struct parities {
ali@56	1813	int dquote,squote;
ali@56	1814	};
ali@56	1815
ali@56	1816	/*
ali@56	1817	* check_for_misspaced_punctuation:
ali@56	1818	*
ali@56	1819	* Look for added or missing spaces around punctuation and quotes.
ali@56	1820	* If there is a punctuation character like ! with no space on
ali@56	1821	* either side, suspect a missing!space. If there are spaces on
ali@56	1822	* both sides , assume a typo. If we see a double quote with no
ali@56	1823	* space or punctuation on either side of it, assume unspaced
ali@56	1824	* quotes "like"this.
ali@56	1825	*/
ali@56	1826	void check_for_misspaced_punctuation(const char *aline,
ali@56	1827	struct parities *parities,int isemptyline)
ali@56	1828	{
ali@56	1829	int i,llen,isacro,isellipsis;
ali@56	1830	const char *s;
ali@56	1831	llen=strlen(aline);
ali@56	1832	for (i=1;i<llen;i++)
ali@56	1833	{
ali@56	1834	/* For each character in the line after the first. */
ali@56	1835	if (strchr(".?!,;:_",aline[i])) /* if it's punctuation */
ali@56	1836	{
ali@56	1837	/* we need to suppress warnings for acronyms like M.D. */
ali@56	1838	isacro=0;
ali@56	1839	/* we need to suppress warnings for ellipsis . . . */
ali@56	1840	isellipsis=0;
ali@56	1841	/* if there are letters on both sides of it or ... */
ali@56	1842	if (gcisalpha(aline[i-1]) && gcisalpha(aline[i+1]) \|\|
ali@56	1843	gcisalpha(aline[i+1]) && strchr("?!,;:",aline[i]))
ali@56	1844	{
ali@56	1845	/* ...if it's strict punctuation followed by an alpha */
ali@56	1846	if (aline[i]=='.')
ali@56	1847	{
ali@56	1848	if (i>2 && aline[i-2]=='.')
ali@56	1849	isacro=1;
ali@56	1850	if (i+2<llen && aline[i+2]=='.')
ali@56	1851	isacro=1;
ali@56	1852	}
ali@56	1853	if (!isacro)
ali@56	1854	{
ali@56	1855	if (pswit[ECHO_SWITCH])
ali@56	1856	printf("\n%s\n",aline);
ali@56	1857	if (!pswit[OVERVIEW_SWITCH])
ali@56	1858	printf(" Line %ld column %d - Missing space?\n",
ali@56	1859	linecnt,i+1);
ali@56	1860	else
ali@56	1861	cnt_punct++;
ali@56	1862	}
ali@56	1863	}
ali@56	1864	if (aline[i-1]==CHAR_SPACE &&
ali@56	1865	(aline[i+1]==CHAR_SPACE \|\| aline[i+1]==0))
ali@56	1866	{
ali@56	1867	/*
ali@56	1868	* If there are spaces on both sides,
ali@56	1869	* or space before and end of line.
ali@56	1870	*/
ali@56	1871	if (aline[i]=='.')
ali@56	1872	{
ali@56	1873	if (i>2 && aline[i-2]=='.')
ali@56	1874	isellipsis=1;
ali@56	1875	if (i+2<llen && aline[i+2]=='.')
ali@56	1876	isellipsis=1;
ali@56	1877	}
ali@56	1878	if (!isemptyline && !isellipsis)
ali@56	1879	{
ali@56	1880	if (pswit[ECHO_SWITCH])
ali@56	1881	printf("\n%s\n",aline);
ali@56	1882	if (!pswit[OVERVIEW_SWITCH])
ali@56	1883	printf(" Line %ld column %d - "
ali@56	1884	"Spaced punctuation?\n",linecnt,i+1);
ali@56	1885	else
ali@56	1886	cnt_punct++;
ali@56	1887	}
ali@56	1888	}
ali@56	1889	}
ali@56	1890	}
ali@56	1891	/* Split out the characters that CANNOT be preceded by space. */
ali@56	1892	llen=strlen(aline);
ali@56	1893	for (i=1;i<llen;i++)
ali@56	1894	{
ali@56	1895	/* for each character in the line after the first */
ali@56	1896	if (strchr("?!,;:",aline[i]))
ali@56	1897	{
ali@56	1898	/* if it's punctuation that _cannot_ have a space before it */
ali@56	1899	if (aline[i-1]==CHAR_SPACE && !isemptyline &&
ali@56	1900	aline[i+1]!=CHAR_SPACE)
ali@56	1901	{
ali@56	1902	/*
ali@56	1903	* If aline[i+1) DOES == space,
ali@56	1904	* it was already reported just above.
ali@56	1905	*/
ali@56	1906	if (pswit[ECHO_SWITCH])
ali@56	1907	printf("\n%s\n",aline);
ali@56	1908	if (!pswit[OVERVIEW_SWITCH])
ali@56	1909	printf(" Line %ld column %d - Spaced punctuation?\n",
ali@56	1910	linecnt,i+1);
ali@56	1911	else
ali@56	1912	cnt_punct++;
ali@56	1913	}
ali@56	1914	}
ali@56	1915	}
ali@56	1916	/*
ali@56	1917	* Special case " .X" where X is any alpha.
ali@56	1918	* This plugs a hole in the acronym code above.
ali@56	1919	* Inelegant, but maintainable.
ali@56	1920	*/
ali@56	1921	llen=strlen(aline);
ali@56	1922	for (i=1;i<llen;i++)
ali@56	1923	{
ali@56	1924	/* for each character in the line after the first */
ali@56	1925	if (aline[i]=='.')
ali@56	1926	{
ali@56	1927	/* if it's a period */
ali@56	1928	if (aline[i-1]==CHAR_SPACE && gcisalpha(aline[i+1]))
ali@56	1929	{
ali@56	1930	/*
ali@56	1931	* If the period follows a space and
ali@56	1932	* is followed by a letter.
ali@56	1933	*/
ali@56	1934	if (pswit[ECHO_SWITCH])
ali@56	1935	printf("\n%s\n",aline);
ali@56	1936	if (!pswit[OVERVIEW_SWITCH])
ali@56	1937	printf(" Line %ld column %d - Spaced punctuation?\n",
ali@56	1938	linecnt,i+1);
ali@56	1939	else
ali@56	1940	cnt_punct++;
ali@56	1941	}
ali@56	1942	}
ali@56	1943	}
ali@56	1944	for (i=1;i<llen;i++)
ali@56	1945	{
ali@56	1946	/* for each character in the line after the first */
ali@56	1947	if (aline[i]==CHAR_DQUOTE)
ali@56	1948	{
ali@56	1949	if (!strchr(" _-.'`,;:!/([{?}])",aline[i-1]) &&
ali@56	1950	!strchr(" _-.'`,;:!/([{?}])",aline[i+1]) && aline[i+1] \|\|
ali@56	1951	!strchr(" _-([{'`",aline[i-1]) && gcisalpha(aline[i+1]))
ali@56	1952	{
ali@56	1953	if (pswit[ECHO_SWITCH])
ali@56	1954	printf("\n%s\n",aline);
ali@56	1955	if (!pswit[OVERVIEW_SWITCH])
ali@56	1956	printf(" Line %ld column %d - Unspaced quotes?\n",
ali@56	1957	linecnt,i+1);
ali@56	1958	else
ali@56	1959	cnt_punct++;
ali@56	1960	}
ali@56	1961	}
ali@56	1962	}
ali@56	1963	/* Check parity of quotes. */
ali@56	1964	for (s=aline;*s;s++)
ali@56	1965	{
ali@56	1966	if (*s==CHAR_DQUOTE)
ali@56	1967	{
ali@56	1968	parities->dquote=!parities->dquote;
ali@56	1969	if (!parities->dquote)
ali@56	1970	{
ali@56	1971	/* parity even */
ali@56	1972	if (!strchr("_-.'`/,;:!?)]} ",s[1]))
ali@56	1973	{
ali@56	1974	if (pswit[ECHO_SWITCH])
ali@56	1975	printf("\n%s\n",aline);
ali@56	1976	if (!pswit[OVERVIEW_SWITCH])
ali@56	1977	printf(" Line %ld column %d - "
ali@56	1978	"Wrongspaced quotes?\n",linecnt,(int)(s-aline)+1);
ali@56	1979	else
ali@56	1980	cnt_punct++;
ali@56	1981	}
ali@56	1982	}
ali@56	1983	else
ali@56	1984	{
ali@56	1985	/* parity odd */
ali@56	1986	if (!gcisalpha(s[1]) && !isdigit(s[1]) &&
ali@56	1987	!strchr("_-/.'`([{$",s[1]) \|\| !s[1])
ali@56	1988	{
ali@56	1989	if (pswit[ECHO_SWITCH])
ali@56	1990	printf("\n%s\n",aline);
ali@56	1991	if (!pswit[OVERVIEW_SWITCH])
ali@56	1992	printf(" Line %ld column %d - "
ali@56	1993	"Wrongspaced quotes?\n",linecnt,(int)(s-aline)+1);
ali@56	1994	else
ali@56	1995	cnt_punct++;
ali@56	1996	}
ali@56	1997	}
ali@56	1998	}
ali@56	1999	}
ali@56	2000	if (*aline==CHAR_DQUOTE)
ali@56	2001	{
ali@56	2002	if (strchr(",;:!?)]} ",aline[1]))
ali@56	2003	{
ali@56	2004	if (pswit[ECHO_SWITCH])
ali@56	2005	printf("\n%s\n",aline);
ali@56	2006	if (!pswit[OVERVIEW_SWITCH])
ali@56	2007	printf(" Line %ld column 1 - Wrongspaced quotes?\n",
ali@56	2008	linecnt);
ali@56	2009	else
ali@56	2010	cnt_punct++;
ali@56	2011	}
ali@56	2012	}
ali@56	2013	if (pswit[SQUOTE_SWITCH])
ali@56	2014	{
ali@56	2015	for (s=aline;*s;s++)
ali@56	2016	{
ali@56	2017	if ((s==CHAR_SQUOTE \|\| s==CHAR_OPEN_SQUOTE) &&
ali@56	2018	(s==aline \|\| s>aline && !gcisalpha(s[-1]) \|\|
ali@56	2019	!gcisalpha(s[1])))
ali@56	2020	{
ali@56	2021	parities->squote=!parities->squote;
ali@56	2022	if (!parities->squote)
ali@56	2023	{
ali@56	2024	/* parity even */
ali@56	2025	if (!strchr("_-.'`/\",;:!?)]} ",s[1]))
ali@56	2026	{
ali@56	2027	if (pswit[ECHO_SWITCH])
ali@56	2028	printf("\n%s\n",aline);
ali@56	2029	if (!pswit[OVERVIEW_SWITCH])
ali@56	2030	printf(" Line %ld column %d - "
ali@56	2031	"Wrongspaced singlequotes?\n",
ali@56	2032	linecnt,(int)(s-aline)+1);
ali@56	2033	else
ali@56	2034	cnt_punct++;
ali@56	2035	}
ali@56	2036	}
ali@56	2037	else
ali@56	2038	{
ali@56	2039	/* parity odd */
ali@56	2040	if (!gcisalpha(s[1]) && !isdigit(s[1]) &&
ali@56	2041	!strchr("_-/\".'`",s[1]) \|\| !s[1])
ali@56	2042	{
ali@56	2043	if (pswit[ECHO_SWITCH])
ali@56	2044	printf("\n%s\n",aline);
ali@56	2045	if (!pswit[OVERVIEW_SWITCH])
ali@56	2046	printf(" Line %ld column %d - "
ali@56	2047	"Wrongspaced singlequotes?\n",
ali@56	2048	linecnt,(int)(s-aline)+1);
ali@56	2049	else
ali@56	2050	cnt_punct++;
ali@56	2051	}
ali@56	2052	}
ali@56	2053	}
ali@56	2054	}
ali@56	2055	}
ali@56	2056	}
ali@56	2057
ali@55	2058	/*
ali@57	2059	* check_for_double_punctuation:
ali@57	2060	*
ali@57	2061	* Look for double punctuation like ,. or ,,
ali@57	2062	* Thanks to DW for the suggestion!
ali@57	2063	* In books with references, ".," and ".;" are common
ali@57	2064	* e.g. "etc., etc.," and vol. 1.; vol 3.;
ali@57	2065	* OTOH, from my initial tests, there are also fairly
ali@57	2066	* common errors. What to do? Make these cases paranoid?
ali@57	2067	* ".," is the most common, so warnings->dotcomma is used
ali@57	2068	* to suppress detailed reporting if it occurs often.
ali@57	2069	*/
ali@57	2070	void check_for_double_punctuation(const char aline,struct warnings warnings)
ali@57	2071	{
ali@57	2072	int i,llen;
ali@57	2073	llen=strlen(aline);
ali@57	2074	for (i=0;i<llen;i++)
ali@57	2075	{
ali@57	2076	/* for each punctuation character in the line */
ali@57	2077	if (strchr(".?!,;:",aline[i]) && strchr(".?!,;:",aline[i+1]) &&
ali@57	2078	aline[i] && aline[i+1])
ali@57	2079	{
ali@57	2080	/* followed by punctuation, it's a query, unless . . . */
ali@57	2081	if (aline[i]==aline[i+1] && (aline[i]=='.' \|\| aline[i]=='?' \|\|
ali@57	2082	aline[i]=='!') \|\|
ali@57	2083	!warnings->dotcomma && aline[i]=='.' && aline[i+1]==',' \|\|
ali@57	2084	warnings->isFrench && !strncmp(aline+i,",...",4) \|\|
ali@57	2085	warnings->isFrench && !strncmp(aline+i,"...,",4) \|\|
ali@57	2086	warnings->isFrench && !strncmp(aline+i,";...",4) \|\|
ali@57	2087	warnings->isFrench && !strncmp(aline+i,"...;",4) \|\|
ali@57	2088	warnings->isFrench && !strncmp(aline+i,":...",4) \|\|
ali@57	2089	warnings->isFrench && !strncmp(aline+i,"...:",4) \|\|
ali@57	2090	warnings->isFrench && !strncmp(aline+i,"!...",4) \|\|
ali@57	2091	warnings->isFrench && !strncmp(aline+i,"...!",4) \|\|
ali@57	2092	warnings->isFrench && !strncmp(aline+i,"?...",4) \|\|
ali@57	2093	warnings->isFrench && !strncmp(aline+i,"...?",4))
ali@57	2094	{
ali@57	2095	if (warnings->isFrench && !strncmp(aline+i,",...",4) \|\|
ali@57	2096	warnings->isFrench && !strncmp(aline+i,"...,",4) \|\|
ali@57	2097	warnings->isFrench && !strncmp(aline+i,";...",4) \|\|
ali@57	2098	warnings->isFrench && !strncmp(aline+i,"...;",4) \|\|
ali@57	2099	warnings->isFrench && !strncmp(aline+i,":...",4) \|\|
ali@57	2100	warnings->isFrench && !strncmp(aline+i,"...:",4) \|\|
ali@57	2101	warnings->isFrench && !strncmp(aline+i,"!...",4) \|\|
ali@57	2102	warnings->isFrench && !strncmp(aline+i,"...!",4) \|\|
ali@57	2103	warnings->isFrench && !strncmp(aline+i,"?...",4) \|\|
ali@57	2104	warnings->isFrench && !strncmp(aline+i,"...?",4))
ali@57	2105	i+=4;
ali@57	2106	; /* do nothing for .. !! and ?? which can be legit */
ali@57	2107	}
ali@57	2108	else
ali@57	2109	{
ali@57	2110	if (pswit[ECHO_SWITCH])
ali@57	2111	printf("\n%s\n",aline);
ali@57	2112	if (!pswit[OVERVIEW_SWITCH])
ali@57	2113	printf(" Line %ld column %d - Double punctuation?\n",
ali@57	2114	linecnt,i+1);
ali@57	2115	else
ali@57	2116	cnt_punct++;
ali@57	2117	}
ali@57	2118	}
ali@57	2119	}
ali@57	2120	}
ali@57	2121
ali@57	2122	/*
ali@58	2123	* check_for_spaced_quotes:
ali@58	2124	*/
ali@58	2125	void check_for_spaced_quotes(const char *aline)
ali@58	2126	{
ali@58	2127	const char s,t;
ali@58	2128	s=aline;
ali@58	2129	while ((t=strstr(s," \" ")))
ali@58	2130	{
ali@58	2131	if (pswit[ECHO_SWITCH])
ali@58	2132	printf("\n%s\n",aline);
ali@58	2133	if (!pswit[OVERVIEW_SWITCH])
ali@58	2134	printf(" Line %ld column %d - Spaced doublequote?\n",
ali@58	2135	linecnt,(int)(t-aline+1));
ali@58	2136	else
ali@58	2137	cnt_punct++;
ali@58	2138	s=t+2;
ali@58	2139	}
ali@58	2140	s=aline;
ali@58	2141	while ((t=strstr(s," ' ")))
ali@58	2142	{
ali@58	2143	if (pswit[ECHO_SWITCH])
ali@58	2144	printf("\n%s\n",aline);
ali@58	2145	if (!pswit[OVERVIEW_SWITCH])
ali@58	2146	printf(" Line %ld column %d - Spaced singlequote?\n",
ali@58	2147	linecnt,(int)(t-aline+1));
ali@58	2148	else
ali@58	2149	cnt_punct++;
ali@58	2150	s=t+2;
ali@58	2151	}
ali@58	2152	s=aline;
ali@58	2153	while ((t=strstr(s," ` ")))
ali@58	2154	{
ali@58	2155	if (pswit[ECHO_SWITCH])
ali@58	2156	printf("\n%s\n",aline);
ali@58	2157	if (!pswit[OVERVIEW_SWITCH])
ali@58	2158	printf(" Line %ld column %d - Spaced singlequote?\n",
ali@58	2159	linecnt,(int)(t-aline+1));
ali@58	2160	else
ali@58	2161	cnt_punct++;
ali@58	2162	s=t+2;
ali@58	2163	}
ali@58	2164	}
ali@58	2165
ali@58	2166	/*
ali@59	2167	* check_for_miscased_genative:
ali@59	2168	*
ali@59	2169	* Check special case of 'S instead of 's at end of word.
ali@59	2170	*/
ali@59	2171	void check_for_miscased_genative(const char *aline)
ali@59	2172	{
ali@59	2173	const char *s;
ali@59	2174	s=aline+1;
ali@59	2175	while (*s)
ali@59	2176	{
ali@59	2177	if (*s==CHAR_SQUOTE && s[1]=='S' && s[-1]>='a' && s[-1]<='z')
ali@59	2178	{
ali@59	2179	if (pswit[ECHO_SWITCH])
ali@59	2180	printf("\n%s\n",aline);
ali@59	2181	if (!pswit[OVERVIEW_SWITCH])
ali@59	2182	printf(" Line %ld column %d - Capital \"S\"?\n",
ali@59	2183	linecnt,(int)(s-aline+2));
ali@59	2184	else
ali@59	2185	cnt_punct++;
ali@59	2186	}
ali@59	2187	s++;
ali@59	2188	}
ali@59	2189	}
ali@59	2190
ali@59	2191	/*
ali@60	2192	* check_end_of_line:
ali@60	2193	*
ali@60	2194	* Now check special cases - start and end of line -
ali@60	2195	* for single and double quotes. Start is sometimes [sic]
ali@60	2196	* but better to query it anyway.
ali@60	2197	* While we're here, check for dash at end of line.
ali@60	2198	*/
ali@60	2199	void check_end_of_line(const char aline,struct warnings warnings)
ali@60	2200	{
ali@60	2201	int i,llen;
ali@60	2202	llen=strlen(aline);
ali@60	2203	if (llen>1)
ali@60	2204	{
ali@60	2205	if (aline[llen-1]==CHAR_DQUOTE \|\| aline[llen-1]==CHAR_SQUOTE \|\|
ali@60	2206	aline[llen-1]==CHAR_OPEN_SQUOTE)
ali@60	2207	if (aline[llen-2]==CHAR_SPACE)
ali@60	2208	{
ali@60	2209	if (pswit[ECHO_SWITCH])
ali@60	2210	printf("\n%s\n",aline);
ali@60	2211	if (!pswit[OVERVIEW_SWITCH])
ali@60	2212	printf(" Line %ld column %d - Spaced quote?\n",
ali@60	2213	linecnt,llen);
ali@60	2214	else
ali@60	2215	cnt_punct++;
ali@60	2216	}
ali@60	2217	if ((aline[0]==CHAR_SQUOTE \|\| aline[0]==CHAR_OPEN_SQUOTE) &&
ali@60	2218	aline[1]==CHAR_SPACE)
ali@60	2219	{
ali@60	2220	if (pswit[ECHO_SWITCH])
ali@60	2221	printf("\n%s\n",aline);
ali@60	2222	if (!pswit[OVERVIEW_SWITCH])
ali@60	2223	printf(" Line %ld column 1 - Spaced quote?\n",linecnt);
ali@60	2224	else
ali@60	2225	cnt_punct++;
ali@60	2226	}
ali@60	2227	/*
ali@60	2228	* Dash at end of line may well be legit - paranoid mode only
ali@60	2229	* and don't report em-dash at line-end.
ali@60	2230	*/
ali@60	2231	if (pswit[PARANOID_SWITCH] && warnings->hyphen)
ali@60	2232	{
ali@60	2233	for (i=llen-1;i>0 && (unsigned char)aline[i]<=CHAR_SPACE;i--)
ali@60	2234	;
ali@60	2235	if (aline[i]=='-' && aline[i-1]!='-')
ali@60	2236	{
ali@60	2237	if (pswit[ECHO_SWITCH])
ali@60	2238	printf("\n%s\n",aline);
ali@60	2239	if (!pswit[OVERVIEW_SWITCH])
ali@60	2240	printf(" Line %ld column %d - Hyphen at end of line?\n",
ali@60	2241	linecnt,i);
ali@60	2242	}
ali@60	2243	}
ali@60	2244	}
ali@60	2245	}
ali@60	2246
ali@60	2247	/*
ali@61	2248	* check_for_unspaced_bracket:
ali@61	2249	*
ali@61	2250	* Brackets are often unspaced, but shouldn't be surrounded by alpha.
ali@61	2251	* If so, suspect a scanno like "a]most".
ali@61	2252	*/
ali@61	2253	void check_for_unspaced_bracket(const char *aline)
ali@61	2254	{
ali@61	2255	int i,llen;
ali@61	2256	llen=strlen(aline);
ali@61	2257	for (i=1;i<llen-1;i++)
ali@61	2258	{
ali@61	2259	/* for each bracket character in the line except 1st & last */
ali@61	2260	if (strchr("{[()]}",aline[i]) && gcisalpha(aline[i-1]) &&
ali@61	2261	gcisalpha(aline[i+1]))
ali@61	2262	{
ali@61	2263	if (pswit[ECHO_SWITCH])
ali@61	2264	printf("\n%s\n",aline);
ali@61	2265	if (!pswit[OVERVIEW_SWITCH])
ali@61	2266	printf(" Line %ld column %d - Unspaced bracket?\n",
ali@61	2267	linecnt,i);
ali@61	2268	else
ali@61	2269	cnt_punct++;
ali@61	2270	}
ali@61	2271	}
ali@61	2272	}
ali@61	2273
ali@61	2274	/*
ali@62	2275	* check_for_unpunctuated_endquote:
ali@62	2276	*/
ali@62	2277	void check_for_unpunctuated_endquote(const char *aline)
ali@62	2278	{
ali@62	2279	int i,llen;
ali@62	2280	llen=strlen(aline);
ali@62	2281	for (i=1;i<llen;i++)
ali@62	2282	{
ali@62	2283	/* for each character in the line except 1st */
ali@62	2284	if (aline[i]==CHAR_DQUOTE && isalpha(aline[i-1]))
ali@62	2285	{
ali@62	2286	if (pswit[ECHO_SWITCH])
ali@62	2287	printf("\n%s\n",aline);
ali@62	2288	if (!pswit[OVERVIEW_SWITCH])
ali@62	2289	printf(" Line %ld column %d - "
ali@62	2290	"endquote missing punctuation?\n",linecnt,i);
ali@62	2291	else
ali@62	2292	cnt_punct++;
ali@62	2293	}
ali@62	2294	}
ali@62	2295	}
ali@62	2296
ali@62	2297	/*
ali@63	2298	* check_for_html_tag:
ali@63	2299	*
ali@63	2300	* Check for <HTML TAG>.
ali@63	2301	*
ali@63	2302	* If there is a < in the line, followed at some point
ali@63	2303	* by a > then we suspect HTML.
ali@63	2304	*/
ali@63	2305	void check_for_html_tag(const char *aline)
ali@63	2306	{
ali@63	2307	int i;
ali@63	2308	const char open,close;
ali@63	2309	open=strstr(aline,"<");
ali@63	2310	if (open)
ali@63	2311	{
ali@63	2312	close=strstr(aline,">");
ali@63	2313	if (close)
ali@63	2314	{
ali@63	2315	i=(signed int)(close-open+1);
ali@63	2316	if (i>0)
ali@63	2317	{
ali@63	2318	strncpy(wrk,open,i);
ali@63	2319	wrk[i]=0;
ali@63	2320	if (pswit[ECHO_SWITCH])
ali@63	2321	printf("\n%s\n",aline);
ali@63	2322	if (!pswit[OVERVIEW_SWITCH])
ali@63	2323	printf(" Line %ld column %d - HTML Tag? %s \n",
ali@63	2324	linecnt,(int)(open-aline)+1,wrk);
ali@63	2325	else
ali@63	2326	cnt_html++;
ali@63	2327	}
ali@63	2328	}
ali@63	2329	}
ali@63	2330	}
ali@63	2331
ali@63	2332	/*
ali@64	2333	* check_for_html_entity:
ali@64	2334	*
ali@64	2335	* Check for &symbol; HTML.
ali@64	2336	*
ali@64	2337	* If there is a & in the line, followed at
ali@64	2338	* some point by a ; then we suspect HTML.
ali@64	2339	*/
ali@64	2340	void check_for_html_entity(const char *aline)
ali@64	2341	{
ali@64	2342	int i;
ali@64	2343	const char s,amp,*scolon;
ali@64	2344	amp=strstr(aline,"&");
ali@64	2345	if (amp)
ali@64	2346	{
ali@64	2347	scolon=strstr(aline,";");
ali@64	2348	if (scolon)
ali@64	2349	{
ali@64	2350	i=(int)(scolon-amp+1);
ali@64	2351	for (s=amp;s<scolon;s++)
ali@64	2352	if (*s==CHAR_SPACE)
ali@64	2353	i=0; /* Don't report "Jones & Son;" */
ali@64	2354	if (i>0)
ali@64	2355	{
ali@64	2356	strncpy(wrk,amp,i);
ali@64	2357	wrk[i]=0;
ali@64	2358	if (pswit[ECHO_SWITCH])
ali@64	2359	printf("\n%s\n",aline);
ali@64	2360	if (!pswit[OVERVIEW_SWITCH])
ali@64	2361	printf(" Line %ld column %d - HTML symbol? %s \n",
ali@64	2362	linecnt,(int)(amp-aline)+1,wrk);
ali@64	2363	else
ali@64	2364	cnt_html++;
ali@64	2365	}
ali@64	2366	}
ali@64	2367	}
ali@64	2368	}
ali@64	2369
ali@65	2370	struct pending {
ali@65	2371	char dquote[80],squote[80],rbrack[80],sbrack[80],cbrack[80],unders[80];
ali@65	2372	long squot;
ali@65	2373	};
ali@65	2374
ali@65	2375	/*
ali@65	2376	* print_pending:
ali@65	2377	*
ali@65	2378	* If we are in a state of unbalanced quotes, and this line
ali@65	2379	* doesn't begin with a quote, output the stored error message.
ali@65	2380	* If the -P switch was used, print the warning even if the
ali@65	2381	* new para starts with quotes.
ali@65	2382	*/
ali@65	2383	void print_pending(const char aline,const char parastart,
ali@65	2384	struct pending *pending)
ali@65	2385	{
ali@65	2386	const char *s;
ali@65	2387	s=aline;
ali@65	2388	while (*s==' ')
ali@65	2389	s++;
ali@65	2390	if (*pending->dquote)
ali@65	2391	if (*s!=CHAR_DQUOTE \|\| pswit[QPARA_SWITCH])
ali@65	2392	{
ali@65	2393	if (!pswit[OVERVIEW_SWITCH])
ali@65	2394	{
ali@65	2395	if (pswit[ECHO_SWITCH])
ali@65	2396	printf("\n%s\n",parastart);
ali@65	2397	puts(pending->dquote);
ali@65	2398	}
ali@65	2399	else
ali@65	2400	cnt_dquot++;
ali@65	2401	}
ali@65	2402	if (*pending->squote)
ali@65	2403	{
ali@65	2404	if (s!=CHAR_SQUOTE && s!=CHAR_OPEN_SQUOTE \|\| pswit[QPARA_SWITCH] \|\|
ali@65	2405	pending->squot)
ali@65	2406	{
ali@65	2407	if (!pswit[OVERVIEW_SWITCH])
ali@65	2408	{
ali@65	2409	if (pswit[ECHO_SWITCH])
ali@65	2410	printf("\n%s\n",parastart);
ali@65	2411	puts(pending->squote);
ali@65	2412	}
ali@65	2413	else
ali@65	2414	cnt_squot++;
ali@65	2415	}
ali@65	2416	}
ali@65	2417	if (*pending->rbrack)
ali@65	2418	{
ali@65	2419	if (!pswit[OVERVIEW_SWITCH])
ali@65	2420	{
ali@65	2421	if (pswit[ECHO_SWITCH])
ali@65	2422	printf("\n%s\n",parastart);
ali@65	2423	puts(pending->rbrack);
ali@65	2424	}
ali@65	2425	else
ali@65	2426	cnt_brack++;
ali@65	2427	}
ali@65	2428	if (*pending->sbrack)
ali@65	2429	{
ali@65	2430	if (!pswit[OVERVIEW_SWITCH])
ali@65	2431	{
ali@65	2432	if (pswit[ECHO_SWITCH])
ali@65	2433	printf("\n%s\n",parastart);
ali@65	2434	puts(pending->sbrack);
ali@65	2435	}
ali@65	2436	else
ali@65	2437	cnt_brack++;
ali@65	2438	}
ali@65	2439	if (*pending->cbrack)
ali@65	2440	{
ali@65	2441	if (!pswit[OVERVIEW_SWITCH])
ali@65	2442	{
ali@65	2443	if (pswit[ECHO_SWITCH])
ali@65	2444	printf("\n%s\n",parastart);
ali@65	2445	puts(pending->cbrack);
ali@65	2446	}
ali@65	2447	else
ali@65	2448	cnt_brack++;
ali@65	2449	}
ali@65	2450	if (*pending->unders)
ali@65	2451	{
ali@65	2452	if (!pswit[OVERVIEW_SWITCH])
ali@65	2453	{
ali@65	2454	if (pswit[ECHO_SWITCH])
ali@65	2455	printf("\n%s\n",parastart);
ali@65	2456	puts(pending->unders);
ali@65	2457	}
ali@65	2458	else
ali@65	2459	cnt_brack++;
ali@65	2460	}
ali@65	2461	}
ali@65	2462
ali@65	2463	/*
ali@65	2464	* check_for_mismatched_quotes:
ali@65	2465	*
ali@65	2466	* At end of paragraph, check for mismatched quotes.
ali@65	2467	*
ali@65	2468	* We don't want to report an error immediately, since it is a
ali@65	2469	* common convention to omit the quotes at end of paragraph if
ali@65	2470	* the next paragraph is a continuation of the same speaker.
ali@65	2471	* Where this is the case, the next para should begin with a
ali@65	2472	* quote, so we store the warning message and only display it
ali@65	2473	* at the top of the next iteration if the new para doesn't
ali@65	2474	* start with a quote.
ali@65	2475	* The -p switch overrides this default, and warns of unclosed
ali@65	2476	* quotes on _every_ paragraph, whether the next begins with a
ali@65	2477	* quote or not.
ali@65	2478	*/
ali@65	2479	void check_for_mismatched_quotes(const struct counters *counters,
ali@65	2480	struct pending *pending)
ali@65	2481	{
ali@65	2482	if (counters->quot%2)
ali@65	2483	sprintf(pending->dquote," Line %ld - Mismatched quotes",
ali@65	2484	linecnt);
ali@65	2485	if (pswit[SQUOTE_SWITCH] && counters->open_single_quote &&
ali@65	2486	counters->open_single_quote!=counters->close_single_quote)
ali@65	2487	sprintf(pending->squote," Line %ld - Mismatched singlequotes?",
ali@65	2488	linecnt);
ali@65	2489	if (pswit[SQUOTE_SWITCH] && counters->open_single_quote &&
ali@65	2490	counters->open_single_quote!=counters->close_single_quote &&
ali@65	2491	counters->open_single_quote!=counters->close_single_quote+1)
ali@65	2492	/*
ali@65	2493	* Flag it to be noted regardless of the
ali@65	2494	* first char of the next para.
ali@65	2495	*/
ali@65	2496	pending->squot=1;
ali@65	2497	if (counters->r_brack)
ali@65	2498	sprintf(pending->rbrack," Line %ld - Mismatched round brackets?",
ali@65	2499	linecnt);
ali@65	2500	if (counters->s_brack)
ali@65	2501	sprintf(pending->sbrack," Line %ld - Mismatched square brackets?",
ali@65	2502	linecnt);
ali@65	2503	if (counters->c_brack)
ali@65	2504	sprintf(pending->cbrack," Line %ld - Mismatched curly brackets?",
ali@65	2505	linecnt);
ali@65	2506	if (counters->c_unders%2)
ali@65	2507	sprintf(pending->unders," Line %ld - Mismatched underscores?",
ali@65	2508	linecnt);
ali@65	2509	}
ali@65	2510
ali@64	2511	/*
ali@66	2512	* check_for_omitted_punctuation:
ali@66	2513	*
ali@66	2514	* Check for omitted punctuation at end of paragraph by working back
ali@66	2515	* through prevline. DW.
ali@66	2516	* Need to check this only for "normal" paras.
ali@66	2517	* So what is a "normal" para?
ali@66	2518	* Not normal if one-liner (chapter headings, etc.)
ali@66	2519	* Not normal if doesn't contain at least one locase letter
ali@66	2520	* Not normal if starts with space
ali@66	2521	*/
ali@66	2522	void check_for_omitted_punctuation(const char *prevline,
ali@66	2523	struct line_properties *last,int start_para_line)
ali@66	2524	{
ali@66	2525	int i;
ali@66	2526	const char *s;
ali@66	2527	for (s=prevline,i=0;*s && !i;s++)
ali@66	2528	if (gcisletter(*s))
ali@66	2529	/* use i to indicate the presence of a letter on the line */
ali@66	2530	i=1;
ali@66	2531	/*
ali@66	2532	* This next "if" is a problem.
ali@66	2533	* If we say "start_para_line <= linecnt - 1", that includes
ali@66	2534	* one-line "paragraphs" like chapter heads. Lotsa false positives.
ali@66	2535	* If we say "start_para_line < linecnt - 1" it doesn't, but then it
ali@66	2536	* misses genuine one-line paragraphs.
ali@66	2537	*/
ali@66	2538	if (i && last->blen>2 && start_para_line<linecnt-1 && *prevline>CHAR_SPACE)
ali@66	2539	{
ali@66	2540	for (i=strlen(prevline)-1;
ali@66	2541	(prevline[i]==CHAR_DQUOTE \|\| prevline[i]==CHAR_SQUOTE) &&
ali@66	2542	prevline[i]>CHAR_SPACE && i>0;
ali@66	2543	i--)
ali@66	2544	;
ali@66	2545	for (;i>0;i--)
ali@66	2546	{
ali@66	2547	if (gcisalpha(prevline[i]))
ali@66	2548	{
ali@66	2549	if (pswit[ECHO_SWITCH])
ali@66	2550	printf("\n%s\n",prevline);
ali@66	2551	if (!pswit[OVERVIEW_SWITCH])
ali@66	2552	printf(" Line %ld column %d - "
ali@66	2553	"No punctuation at para end?\n",
ali@66	2554	linecnt-1,strlen(prevline));
ali@66	2555	else
ali@66	2556	cnt_punct++;
ali@66	2557	break;
ali@66	2558	}
ali@66	2559	if (strchr("-.:!([{?}])",prevline[i]))
ali@66	2560	break;
ali@66	2561	}
ali@66	2562	}
ali@66	2563	}
ali@66	2564
ali@66	2565	/*
ali@41	2566	* procfile:
ali@41	2567	*
ali@41	2568	* Process one file.
ali@41	2569	*/
ali@41	2570	void procfile(char *filename)
ali@41	2571	{
ali@65	2572	const char *s;
ali@41	2573	char parastart[81]; /* first line of current para */
ali@41	2574	FILE *infile;
ali@41	2575	struct first_pass_results *first_pass_results;
ali@42	2576	struct warnings *warnings;
ali@43	2577	struct counters counters={0};
ali@45	2578	struct line_properties last={0};
ali@56	2579	struct parities parities={0};
ali@65	2580	struct pending pending={{0},};
ali@43	2581	int isemptyline;
ali@65	2582	long start_para_line;
ali@55	2583	signed int i,llen,isacro,isellipsis;
ali@55	2584	signed int isnewpara;
ali@41	2585	signed int enddash;
ali@45	2586	last.start=CHAR_SPACE;
ali@65	2587	*prevline=0;
ali@41	2588	linecnt=checked_linecnt=start_para_line=0;
ali@53	2589	i=llen=isacro=isellipsis=0;
ali@55	2590	isnewpara=enddash=0;
ali@41	2591	infile=fopen(filename,"rb");
ali@41	2592	if (!infile)
ali@41	2593	{
ali@41	2594	if (pswit[STDOUT_SWITCH])
ali@41	2595	fprintf(stdout,"bookloupe: cannot open %s\n",filename);
ali@41	2596	else
ali@41	2597	fprintf(stderr,"bookloupe: cannot open %s\n",filename);
ali@41	2598	exit(1);
ali@41	2599	}
ali@41	2600	fprintf(stdout,"\n\nFile: %s\n\n",filename);
ali@41	2601	first_pass_results=first_pass(infile);
ali@42	2602	warnings=report_first_pass(first_pass_results);
ali@40	2603	/*
ali@40	2604	* Here we go with the main pass. Hold onto yer hat!
ali@40	2605	*/
ali@65	2606	rewind(infile);
ali@65	2607	linecnt=0;
ali@40	2608	while (flgets(aline,LINEBUFSIZE-1,infile,linecnt+1))
ali@40	2609	{
ali@0	2610	linecnt++;
ali@40	2611	if (linecnt==1)
ali@40	2612	isnewpara=1;
ali@40	2613	if (pswit[DP_SWITCH] && !strncmp(aline,"-----File: ",11))
ali@40	2614	continue; // skip DP page separators completely
ali@41	2615	if (linecnt<first_pass_results->firstline \|\|
ali@41	2616	(first_pass_results->footerline>0 &&
ali@41	2617	linecnt>first_pass_results->footerline))
ali@40	2618	{
ali@40	2619	if (pswit[HEADER_SWITCH])
ali@40	2620	{
ali@40	2621	if (!strncmp(aline,"Title:",6))
ali@40	2622	printf(" %s\n",aline);
ali@40	2623	if (!strncmp(aline,"Author:",7))
ali@40	2624	printf(" %s\n",aline);
ali@40	2625	if (!strncmp(aline,"Release Date:",13))
ali@40	2626	printf(" %s\n",aline);
ali@40	2627	if (!strncmp(aline,"Edition:",8))
ali@40	2628	printf(" %s\n\n",aline);
ali@40	2629	}
ali@0	2630	continue; /* skip through the header */
ali@40	2631	}
ali@0	2632	checked_linecnt++;
ali@65	2633	print_pending(aline,parastart,&pending);
ali@65	2634	memset(&pending,0,sizeof(pending));
ali@43	2635	isemptyline=analyse_quotes(aline,&counters);
ali@40	2636	if (isnewpara && !isemptyline)
ali@40	2637	{
ali@40	2638	/* This line is the start of a new paragraph. */
ali@40	2639	start_para_line=linecnt;
ali@40	2640	/* Capture its first line in case we want to report it later. */
ali@40	2641	strncpy(parastart,aline,80);
ali@40	2642	parastart[79]=0;
ali@56	2643	memset(&parities,0,sizeof(parities)); /* restart the quote count */
ali@40	2644	s=aline;
ali@40	2645	while (!gcisalpha(s) && !gcisdigit(s) && *s)
ali@40	2646	s++;
ali@40	2647	if (s>='a' && s<='z')
ali@40	2648	{
ali@40	2649	/* and its first letter is lowercase */
ali@40	2650	if (pswit[ECHO_SWITCH])
ali@40	2651	printf("\n%s\n",aline);
ali@0	2652	if (!pswit[OVERVIEW_SWITCH])
ali@40	2653	printf(" Line %ld column %d - "
ali@40	2654	"Paragraph starts with lower-case\n",
ali@40	2655	linecnt,(int)(s-aline)+1);
ali@0	2656	else
ali@0	2657	cnt_punct++;
ali@40	2658	}
ali@40	2659	isnewpara=0; /* Signal the end of new para processing. */
ali@40	2660	}
ali@40	2661	/* Check for an em-dash broken at line end. */
ali@40	2662	if (enddash && *aline=='-')
ali@40	2663	{
ali@40	2664	if (pswit[ECHO_SWITCH])
ali@40	2665	printf("\n%s\n",aline);
ali@0	2666	if (!pswit[OVERVIEW_SWITCH])
ali@40	2667	printf(" Line %ld column 1 - Broken em-dash?\n",linecnt);
ali@0	2668	else
ali@0	2669	cnt_punct++;
ali@40	2670	}
ali@40	2671	enddash=0;
ali@40	2672	for (s=aline+strlen(aline)-1;*s==' ' && s>aline;s--)
ali@40	2673	;
ali@40	2674	if (s>=aline && *s=='-')
ali@40	2675	enddash=1;
ali@67	2676	check_for_control_characters(aline);
ali@42	2677	if (warnings->bin)
ali@44	2678	check_for_odd_characters(aline,warnings,isemptyline);
ali@42	2679	if (warnings->longline)
ali@45	2680	check_for_long_line(aline);
ali@45	2681	if (warnings->shortline)
ali@45	2682	check_for_short_line(aline,&last);
ali@45	2683	last.blen=last.len;
ali@45	2684	last.len=strlen(aline);
ali@45	2685	last.start=aline[0];
ali@46	2686	check_for_starting_punctuation(aline);
ali@42	2687	if (warnings->dash)
ali@40	2688	{
ali@47	2689	check_for_spaced_emdash(aline);
ali@47	2690	check_for_spaced_dash(aline);
ali@40	2691	}
ali@48	2692	check_for_unmarked_paragraphs(aline);
ali@49	2693	check_for_jeebies(aline);
ali@50	2694	check_for_mta_from(aline);
ali@51	2695	check_for_orphan_character(aline);
ali@52	2696	check_for_pling_scanno(aline);
ali@53	2697	check_for_extra_period(aline,warnings);
ali@54	2698	check_for_following_punctuation(aline);
ali@55	2699	check_for_typos(aline,warnings);
ali@56	2700	check_for_misspaced_punctuation(aline,&parities,isemptyline);
ali@57	2701	check_for_double_punctuation(aline,warnings);
ali@58	2702	check_for_spaced_quotes(aline);
ali@59	2703	check_for_miscased_genative(aline);
ali@60	2704	check_end_of_line(aline,warnings);
ali@61	2705	check_for_unspaced_bracket(aline);
ali@42	2706	if (warnings->endquote)
ali@62	2707	check_for_unpunctuated_endquote(aline);
ali@63	2708	check_for_html_tag(aline);
ali@64	2709	check_for_html_entity(aline);
ali@40	2710	if (isemptyline)
ali@40	2711	{
ali@65	2712	check_for_mismatched_quotes(&counters,&pending);
ali@43	2713	memset(&counters,0,sizeof(counters));
ali@40	2714	/* let the next iteration know that it's starting a new para */
ali@40	2715	isnewpara=1;
ali@66	2716	check_for_omitted_punctuation(prevline,&last,start_para_line);
ali@40	2717	}
ali@40	2718	strcpy(prevline,aline);
ali@0	2719	}
ali@40	2720	fclose(infile);
ali@0	2721	if (!pswit[OVERVIEW_SWITCH])
ali@40	2722	for (i=0;i<MAX_QWORD;i++)
ali@0	2723	if (dupcnt[i])
ali@40	2724	printf("\nNote: Queried word %s was duplicated %d time%s\n",
ali@40	2725	qword[i],dupcnt[i],"s");
ali@0	2726	}
ali@0	2727
ali@40	2728	/*
ali@40	2729	* flgets:
ali@40	2730	*
ali@40	2731	* Get one line from the input stream, checking for
ali@40	2732	* the existence of exactly one CR/LF line-end per line.
ali@40	2733	*
ali@40	2734	* Returns: a pointer to the line.
ali@40	2735	*/
ali@40	2736	char flgets(char theline,int maxlen,FILE *thefile,long lcnt)
ali@0	2737	{
ali@0	2738	char c;
ali@40	2739	int len,isCR,cint;
ali@40	2740	*theline=0;
ali@40	2741	len=isCR=0;
ali@40	2742	c=cint=fgetc(thefile);
ali@40	2743	do
ali@40	2744	{
ali@40	2745	if (cint==EOF)
ali@40	2746	return NULL;
ali@40	2747	/* either way, it's end of line */
ali@40	2748	if (c==10)
ali@40	2749	{
ali@0	2750	if (isCR)
ali@0	2751	break;
ali@40	2752	else
ali@40	2753	{
ali@40	2754	/* Error - a LF without a preceding CR */
ali@40	2755	if (pswit[LINE_END_SWITCH])
ali@40	2756	{
ali@40	2757	if (pswit[ECHO_SWITCH])
ali@40	2758	printf("\n%s\n",theline);
ali@0	2759	if (!pswit[OVERVIEW_SWITCH])
ali@40	2760	printf(" Line %ld - No CR?\n",lcnt);
ali@0	2761	else
ali@0	2762	cnt_lineend++;
ali@40	2763	}
ali@0	2764	break;
ali@40	2765	}
ali@40	2766	}
ali@40	2767	if (c==13)
ali@40	2768	{
ali@40	2769	if (isCR)
ali@40	2770	{
ali@40	2771	/* Error - two successive CRs */
ali@40	2772	if (pswit[LINE_END_SWITCH])
ali@40	2773	{
ali@40	2774	if (pswit[ECHO_SWITCH])
ali@40	2775	printf("\n%s\n",theline);
ali@0	2776	if (!pswit[OVERVIEW_SWITCH])
ali@40	2777	printf(" Line %ld - Two successive CRs?\n",lcnt);
ali@0	2778	else
ali@0	2779	cnt_lineend++;
ali@40	2780	}
ali@40	2781	}
ali@40	2782	isCR=1;
ali@40	2783	}
ali@40	2784	else
ali@40	2785	{
ali@40	2786	if (pswit[LINE_END_SWITCH] && isCR)
ali@40	2787	{
ali@40	2788	if (pswit[ECHO_SWITCH])
ali@40	2789	printf("\n%s\n",theline);
ali@0	2790	if (!pswit[OVERVIEW_SWITCH])
ali@40	2791	printf(" Line %ld column %d - CR without LF?\n",
ali@40	2792	lcnt,len+1);
ali@0	2793	else
ali@0	2794	cnt_lineend++;
ali@40	2795	}
ali@40	2796	theline[len]=c;
ali@40	2797	len++;
ali@40	2798	theline[len]=0;
ali@40	2799	isCR=0;
ali@40	2800	}
ali@40	2801	c=cint=fgetc(thefile);
ali@40	2802	} while(len<maxlen);
ali@0	2803	if (pswit[MARKUP_SWITCH])
ali@0	2804	postprocess_for_HTML(theline);
ali@0	2805	if (pswit[DP_SWITCH])
ali@0	2806	postprocess_for_DP(theline);
ali@40	2807	return theline;
ali@0	2808	}
ali@0	2809
ali@40	2810	/*
ali@40	2811	* mixdigit:
ali@40	2812	*
ali@40	2813	* Takes a "word" as a parameter, and checks whether it
ali@40	2814	* contains a mixture of alpha and digits. Generally, this is an
ali@40	2815	* error, but may not be for cases like 4th or L5 12s. 3d.
ali@40	2816	*
ali@40	2817	* Returns: 0 if no error found, 1 if error.
ali@40	2818	*/
ali@40	2819	int mixdigit(char *checkword)
ali@0	2820	{
ali@40	2821	int wehaveadigit,wehavealetter,firstdigits,query,wl;
ali@0	2822	char *s;
ali@40	2823	wehaveadigit=wehavealetter=query=0;
ali@40	2824	for (s=checkword;*s;s++)
ali@0	2825	if (gcisalpha(*s))
ali@40	2826	wehavealetter=1;
ali@0	2827	else
ali@0	2828	if (gcisdigit(*s))
ali@40	2829	wehaveadigit=1;
ali@40	2830	if (wehaveadigit && wehavealetter)
ali@40	2831	{
ali@40	2832	/* Now exclude common legit cases, like "21st" and "12l. 3s. 11d." */
ali@40	2833	query=1;
ali@40	2834	wl=strlen(checkword);
ali@40	2835	for (firstdigits=0;gcisdigit(checkword[firstdigits]);firstdigits++)
ali@0	2836	;
ali@0	2837	/* digits, ending in st, rd, nd, th of either case */
ali@40	2838	if (firstdigits+2==wl && (matchword(checkword+wl-2,"st") \|\|
ali@40	2839	matchword(checkword+wl-2,"rd") \|\| matchword(checkword+wl-2,"nd") \|\|
ali@40	2840	matchword(checkword+wl-2,"th")))
ali@40	2841	query=0;
ali@40	2842	if (firstdigits+3==wl && (matchword(checkword+wl-3,"sts") \|\|
ali@40	2843	matchword(checkword+wl-3,"rds") \|\| matchword(checkword+wl-3,"nds") \|\|
ali@40	2844	matchword(checkword+wl-3,"ths")))
ali@40	2845	query=0;
ali@40	2846	if (firstdigits+3==wl && (matchword(checkword+wl-4,"stly") \|\|
ali@40	2847	matchword(checkword+wl-4,"rdly") \|\|
ali@40	2848	matchword(checkword+wl-4,"ndly") \|\| matchword(checkword+wl-4,"thly")))
ali@40	2849	query=0;
ali@0	2850	/* digits, ending in l, L, s or d */
ali@40	2851	if (firstdigits+1==wl && (checkword[wl-1]=='l' \|\|
ali@40	2852	checkword[wl-1]=='L' \|\| checkword[wl-1]=='s' \|\| checkword[wl-1]=='d'))
ali@40	2853	query=0;
ali@40	2854	/*
ali@40	2855	* L at the start of a number, representing Britsh pounds, like L500.
ali@40	2856	* This is cute. We know the current word is mixeddigit. If the first
ali@40	2857	* letter is L, there must be at least one digit following. If both
ali@40	2858	* digits and letters follow, we have a genuine error, else we have a
ali@40	2859	* capital L followed by digits, and we accept that as a non-error.
ali@40	2860	*/
ali@40	2861	if (checkword[0]=='L' && !mixdigit(checkword+1))
ali@40	2862	query=0;
ali@40	2863	}
ali@40	2864	return query;
ali@0	2865	}
ali@0	2866
ali@40	2867	/*
ali@40	2868	* getaword:
ali@40	2869	*
ali@40	2870	* Extracts the first/next "word" from the line, and puts
ali@40	2871	* it into "thisword". A word is defined as one English word unit--or
ali@40	2872	* at least that's the aim.
ali@40	2873	*
ali@40	2874	* Returns: a pointer to the position in the line where we will start
ali@40	2875	* looking for the next word.
ali@40	2876	*/
ali@54	2877	const char getaword(const char fromline,char *thisword)
ali@0	2878	{
ali@40	2879	int i,wordlen;
ali@54	2880	const char *s;
ali@40	2881	wordlen=0;
ali@40	2882	for (;!gcisdigit(fromline) && !gcisalpha(fromline) && *fromline;
ali@40	2883	fromline++)
ali@40	2884	;
ali@40	2885	/*
ali@40	2886	* Use a look-ahead to handle exceptions for numbers like 1,000 and 1.35.
ali@40	2887	* Especially yucky is the case of L1,000
ali@40	2888	* This section looks for a pattern of characters including a digit
ali@40	2889	* followed by a comma or period followed by one or more digits.
ali@40	2890	* If found, it returns this whole pattern as a word; otherwise we discard
ali@40	2891	* the results and resume our normal programming.
ali@40	2892	*/
ali@40	2893	s=fromline;
ali@40	2894	for (;(gcisdigit(s) \|\| gcisalpha(s) \|\| s==',' \|\| s=='.') &&
ali@40	2895	wordlen<MAXWORDLEN;s++)
ali@40	2896	{
ali@40	2897	thisword[wordlen]=*s;
ali@0	2898	wordlen++;
ali@40	2899	}
ali@40	2900	thisword[wordlen]=0;
ali@40	2901	for (i=1;i<wordlen-1;i++)
ali@40	2902	{
ali@40	2903	if (thisword[i]=='.' \|\| thisword[i]==',')
ali@40	2904	{
ali@40	2905	if (gcisdigit(thisword[i-1]) && gcisdigit(thisword[i-1]))
ali@40	2906	{
ali@40	2907	fromline=s;
ali@40	2908	return fromline;
ali@40	2909	}
ali@40	2910	}
ali@40	2911	}
ali@0	2912	/* we didn't find a punctuated number - do the regular getword thing */
ali@40	2913	wordlen=0;
ali@40	2914	for (;(gcisdigit(fromline) \|\| gcisalpha(fromline) \|\| *fromline=='\'') &&
ali@40	2915	wordlen<MAXWORDLEN;fromline++)
ali@40	2916	{
ali@40	2917	thisword[wordlen]=*fromline;
ali@0	2918	wordlen++;
ali@40	2919	}
ali@40	2920	thisword[wordlen]=0;
ali@40	2921	return fromline;
ali@0	2922	}
ali@0	2923
ali@40	2924	/*
ali@40	2925	* matchword:
ali@40	2926	*
ali@40	2927	* A case-insensitive string matcher.
ali@40	2928	*/
ali@40	2929	int matchword(char checkfor,char thisword)
ali@0	2930	{
ali@40	2931	unsigned int ismatch,i;
ali@40	2932	if (strlen(checkfor)!=strlen(thisword))
ali@40	2933	return 0;
ali@40	2934	ismatch=1; /* assume a match until we find a difference */
ali@40	2935	for (i=0;i<strlen(checkfor);i++)
ali@40	2936	if (toupper(checkfor[i])!=toupper(thisword[i]))
ali@40	2937	ismatch=0;
ali@40	2938	return ismatch;
ali@0	2939	}
ali@0	2940
ali@40	2941	/*
ali@40	2942	* lowerit:
ali@40	2943	*
ali@40	2944	* Lowercase the line.
ali@40	2945	*/
ali@0	2946
ali@0	2947	void lowerit(char *theline)
ali@0	2948	{
ali@40	2949	for (;*theline;theline++)
ali@40	2950	if (theline>='A' && theline<='Z')
ali@40	2951	*theline+=32;
ali@0	2952	}
ali@0	2953
ali@40	2954	/*
ali@40	2955	* isroman:
ali@40	2956	*
ali@40	2957	* Is this word a Roman Numeral?
ali@40	2958	*
ali@40	2959	* It doesn't actually validate that the number is a valid Roman Numeral--for
ali@40	2960	* example it will pass MXXXXXXXXXX as a valid Roman Numeral, but that's not
ali@40	2961	* what we're here to do. If it passes this, it LOOKS like a Roman numeral.
ali@40	2962	* Anyway, the actual Romans were pretty tolerant of bad arithmetic, or
ali@40	2963	* expressions thereof, except when it came to taxes. Allow any number of M,
ali@40	2964	* an optional D, an optional CM or CD, any number of optional Cs, an optional
ali@40	2965	* XL or an optional XC, an optional IX or IV, an optional V and any number
ali@40	2966	* of optional Is.
ali@40	2967	*/
ali@0	2968	int isroman(char *t)
ali@0	2969	{
ali@0	2970	char *s;
ali@40	2971	if (!t \|\| !*t)
ali@40	2972	return 0;
ali@40	2973	s=t;
ali@40	2974	while (t=='m' && t)
ali@40	2975	t++;
ali@40	2976	if (*t=='d')
ali@40	2977	t++;
ali@40	2978	if (*t=='c' && t[1]=='m')
ali@40	2979	t+=2;
ali@40	2980	if (*t=='c' && t[1]=='d')
ali@40	2981	t+=2;
ali@40	2982	while (t=='c' && t)
ali@40	2983	t++;
ali@40	2984	if (*t=='x' && t[1]=='l')
ali@40	2985	t+=2;
ali@40	2986	if (*t=='x' && t[1]=='c')
ali@40	2987	t+=2;
ali@40	2988	if (*t=='l')
ali@40	2989	t++;
ali@40	2990	while (t=='x' && t)
ali@40	2991	t++;
ali@40	2992	if (*t=='i' && t[1]=='x')
ali@40	2993	t+=2;
ali@40	2994	if (*t=='i' && t[1]=='v')
ali@40	2995	t+=2;
ali@40	2996	if (*t=='v')
ali@40	2997	t++;
ali@40	2998	while (t=='i' && t)
ali@40	2999	t++;
ali@40	3000	return !*t;
ali@0	3001	}
ali@0	3002
ali@40	3003	/*
ali@40	3004	* gcisalpha:
ali@40	3005	*
ali@40	3006	* A version of isalpha() that is somewhat lenient on 8-bit texts.
ali@40	3007	* If we use the standard function, 8-bit accented characters break
ali@40	3008	* words, so that tete with accented characters appears to be two words, "t"
ali@40	3009	* and "t", with 8-bit characters between them. This causes over-reporting of
ali@40	3010	* errors. gcisalpha() recognizes accented letters from the CP1252 (Windows)
ali@40	3011	* and ISO-8859-1 character sets, which are the most common PG 8-bit types.
ali@40	3012	*/
ali@0	3013	int gcisalpha(unsigned char c)
ali@0	3014	{
ali@40	3015	if (c>='a' && c<='z')
ali@40	3016	return 1;
ali@40	3017	if (c>='A' && c<='Z')
ali@40	3018	return 1;
ali@40	3019	if (c<140)
ali@40	3020	return 0;
ali@40	3021	if (c>=192 && c!=208 && c!=215 && c!=222 && c!=240 && c!=247 && c!=254)
ali@40	3022	return 1;
ali@40	3023	if (c==140 \|\| c==142 \|\| c==156 \|\| c==158 \|\| c==159)
ali@40	3024	return 1;
ali@40	3025	return 0;
ali@0	3026	}
ali@0	3027
ali@40	3028	/*
ali@40	3029	* gcisdigit:
ali@40	3030	*
ali@40	3031	* A version of isdigit() that doesn't get confused in 8-bit texts.
ali@40	3032	*/
ali@0	3033	int gcisdigit(unsigned char c)
ali@0	3034	{
ali@40	3035	return c>='0' && c<='9';
ali@0	3036	}
ali@0	3037
ali@40	3038	/*
ali@40	3039	* gcisletter:
ali@40	3040	*
ali@40	3041	* A version of isletter() that doesn't get confused in 8-bit texts.
ali@40	3042	* NB: this is ISO-8891-1-specific.
ali@40	3043	*/
ali@0	3044	int gcisletter(unsigned char c)
ali@0	3045	{
ali@40	3046	return c>='A' && c<='Z' \|\| c>='a' && c<='z' \|\| c>=192;
ali@0	3047	}
ali@0	3048
ali@40	3049	/*
ali@40	3050	* gcstrchr:
ali@40	3051	*
ali@40	3052	* Wraps strchr to return NULL if the character being searched for is zero.
ali@40	3053	*/
ali@40	3054	char gcstrchr(char s,char c)
ali@0	3055	{
ali@40	3056	if (!c)
ali@40	3057	return NULL;
ali@40	3058	return strchr(s,c);
ali@0	3059	}
ali@0	3060
ali@40	3061	/*
ali@40	3062	* postprocess_for_DP:
ali@40	3063	*
ali@40	3064	* Invoked with the -d switch from flgets().
ali@40	3065	* It simply "removes" from the line a hard-coded set of common
ali@40	3066	* DP-specific tags, so that the line passed to the main routine has
ali@40	3067	* been pre-cleaned of DP markup.
ali@40	3068	*/
ali@0	3069	void postprocess_for_DP(char *theline)
ali@0	3070	{
ali@40	3071	char s,t;
ali@0	3072	int i;
ali@0	3073	if (!*theline)
ali@0	3074	return;
ali@40	3075	for (i=0;*DPmarkup[i];i++)
ali@40	3076	{
ali@40	3077	s=strstr(theline,DPmarkup[i]);
ali@40	3078	while (s)
ali@40	3079	{
ali@40	3080	t=s+strlen(DPmarkup[i]);
ali@40	3081	while (*t)
ali@40	3082	{
ali@40	3083	s=t;
ali@40	3084	t++;
ali@40	3085	s++;
ali@40	3086	}
ali@40	3087	*s=0;
ali@40	3088	s=strstr(theline,DPmarkup[i]);
ali@40	3089	}
ali@40	3090	}
ali@0	3091	}
ali@0	3092
ali@40	3093	/*
ali@40	3094	* postprocess_for_HTML:
ali@40	3095	*
ali@40	3096	* Invoked with the -m switch from flgets().
ali@40	3097	* It simply "removes" from the line a hard-coded set of common
ali@40	3098	* HTML tags and "replaces" a hard-coded set of common HTML
ali@40	3099	* entities, so that the line passed to the main routine has
ali@40	3100	* been pre-cleaned of HTML.
ali@40	3101	*/
ali@0	3102	void postprocess_for_HTML(char *theline)
ali@0	3103	{
ali@40	3104	if (strstr(theline,"<") && strstr(theline,">"))
ali@0	3105	while (losemarkup(theline))
ali@0	3106	;
ali@0	3107	while (loseentities(theline))
ali@0	3108	;
ali@0	3109	}
ali@0	3110
ali@0	3111	char losemarkup(char theline)
ali@0	3112	{
ali@40	3113	char s,t;
ali@0	3114	int i;
ali@0	3115	if (!*theline)
ali@40	3116	return NULL;
ali@40	3117	s=strstr(theline,"<");
ali@40	3118	t=strstr(theline,">");
ali@40	3119	if (!s \|\| !t)
ali@40	3120	return NULL;
ali@40	3121	for (i=0;*markup[i];i++)
ali@40	3122	if (!tagcomp(s+1,markup[i]))
ali@40	3123	{
ali@40	3124	if (!t[1])
ali@40	3125	{
ali@40	3126	*s=0;
ali@40	3127	return s;
ali@40	3128	}
ali@40	3129	else if (t>s)
ali@40	3130	{
ali@40	3131	strcpy(s,t+1);
ali@40	3132	return s;
ali@40	3133	}
ali@0	3134	}
ali@40	3135	/* It's an unrecognized <xxx>. */
ali@40	3136	return NULL;
ali@0	3137	}
ali@0	3138
ali@0	3139	char loseentities(char theline)
ali@0	3140	{
ali@0	3141	int i;
ali@40	3142	char s,t;
ali@0	3143	if (!*theline)
ali@40	3144	return NULL;
ali@40	3145	for (i=0;*entities[i].htmlent;i++)
ali@40	3146	{
ali@40	3147	s=strstr(theline,entities[i].htmlent);
ali@40	3148	if (s)
ali@40	3149	{
ali@40	3150	t=malloc((size_t)strlen(s));
ali@40	3151	if (!t)
ali@40	3152	return NULL;
ali@40	3153	strcpy(t,s+strlen(entities[i].htmlent));
ali@40	3154	strcpy(s,entities[i].textent);
ali@40	3155	strcat(s,t);
ali@0	3156	free(t);
ali@40	3157	return theline;
ali@40	3158	}
ali@40	3159	}
ali@40	3160	for (i=0;*entities[i].htmlnum;i++)
ali@40	3161	{
ali@40	3162	s=strstr(theline,entities[i].htmlnum);
ali@40	3163	if (s)
ali@40	3164	{
ali@40	3165	t=malloc((size_t)strlen(s));
ali@40	3166	if (!t)
ali@40	3167	return NULL;
ali@40	3168	strcpy(t,s+strlen(entities[i].htmlnum));
ali@40	3169	strcpy(s,entities[i].textent);
ali@40	3170	strcat(s,t);
ali@0	3171	free(t);
ali@40	3172	return theline;
ali@40	3173	}
ali@40	3174	}
ali@40	3175	return NULL;
ali@0	3176	}
ali@0	3177
ali@40	3178	int tagcomp(char strin,char basetag)
ali@0	3179	{
ali@40	3180	char s,t;
ali@40	3181	s=basetag;
ali@40	3182	t=strin;
ali@40	3183	if (*t=='/')
ali@40	3184	t++; /* ignore a slash */
ali@40	3185	while (s && t)
ali@40	3186	{
ali@40	3187	if (tolower(s)!=tolower(t))
ali@40	3188	return 1;
ali@40	3189	s++;
ali@40	3190	t++;
ali@40	3191	}
ali@40	3192	return 0;
ali@0	3193	}
ali@0	3194
ali@40	3195	void proghelp()
ali@0	3196	{
ali@40	3197	fputs("Bookloupe version " PACKAGE_VERSION ".\n",stderr);
ali@40	3198	fputs("Copyright 2000-2005 Jim Tinsley <jtinsley@pobox.com>.\n",stderr);
ali@40	3199	fputs("Copyright 2012- J. Ali Harlow <ali@juiblex.co.uk>.\n",stderr);
ali@40	3200	fputs("Bookloupe comes wih ABSOLUTELY NO WARRANTY. "
ali@40	3201	"For details, read the file COPYING.\n",stderr);
ali@40	3202	fputs("This is Free Software; "
ali@40	3203	"you may redistribute it under certain conditions (GPL);\n",stderr);
ali@40	3204	fputs("read the file COPYING for details.\n\n",stderr);
ali@40	3205	fputs("Usage is: bookloupe [-setpxloyhud] filename\n",stderr);
ali@40	3206	fputs(" where -s checks single quotes, -e suppresses echoing lines, "
ali@40	3207	"-t checks typos\n",stderr);
ali@40	3208	fputs(" -x (paranoid) switches OFF -t and extra checks, "
ali@40	3209	"-l turns OFF line-end checks\n",stderr);
ali@40	3210	fputs(" -o just displays overview without detail, "
ali@40	3211	"-h echoes header fields\n",stderr);
ali@40	3212	fputs(" -v (verbose) unsuppresses duplicate reporting, "
ali@40	3213	"-m suppresses markup\n",stderr);
ali@0	3214	fputs(" -d ignores DP-specific markup,\n",stderr);
ali@40	3215	fputs(" -u uses a file gutcheck.typ to query user-defined "
ali@40	3216	"possible typos\n",stderr);
ali@40	3217	fputs("Sample usage: bookloupe warpeace.txt \n",stderr);
ali@0	3218	fputs("\n",stderr);
ali@40	3219	fputs("Bookloupe looks for errors in Project Gutenberg(TM) etexts.\n",
ali@40	3220	stderr);
ali@40	3221	fputs("Bookloupe queries anything it thinks shouldn't be in a PG text; "
ali@40	3222	"non-ASCII\n",stderr);
ali@40	3223	fputs("characters like accented letters, "
ali@40	3224	"lines longer than 75 or shorter than 55,\n",stderr);
ali@40	3225	fputs("unbalanced quotes or brackets, "
ali@40	3226	"a variety of badly formatted punctuation, \n",stderr);
ali@40	3227	fputs("HTML tags, some likely typos. "
ali@40	3228	"It is NOT a substitute for human judgement.\n",stderr);
ali@0	3229	fputs("\n",stderr);
ali@0	3230	}

author	ali <ali@juiblex.co.uk>
	Sun May 26 22:43:45 2013 +0100 (2013-05-26)
changeset 67	865063352146
parent 66	a5ef278feb34
child 68	adb087007d08
permissions	-rw-r--r--