Merge pull request #275 from rrthomas/masterv2.3.0

Add -p option to enchant

6 files changed, 88 insertions, 22 deletions

diff --git a/NEWS b/NEWS
index dc0808d..8e13503 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,20 @@
+2.3.0 (June 14, 2021)
+---------------------
+
+Add -p flag to enchant.
+
+Allow personal wordlist APIs (enchant_pwl_*) to take -1 as the length of the
+word, as the enchant_dict_* APIs already allowed.
+
+Add documentation to enchant_provider.h.
+
+Make Aspell the default backend for English locales it supports, as it is
+much quicker than Hunspell in this case and achieves slightly better
+results.
+
+Require nuspell version 4.1.0 or later.
+
+
 2.2.15 (December 22, 2020)
 --------------------------
 
diff --git a/configure.ac b/configure.ac
index 4877edc..e67163e 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1,4 +1,4 @@
-AC_INIT([enchant],[2.2.15])
+AC_INIT([enchant],[2.3.0])
 AC_CONFIG_SRCDIR(src/enchant.h)
 AC_CONFIG_AUX_DIR([build-aux])
 AM_INIT_AUTOMAKE([subdir-objects])
diff --git a/src/enchant.1.in b/src/enchant.1.in
index 3b76c3a..4ee07f6 100644
--- a/src/enchant.1.in
+++ b/src/enchant.1.in
@@ -14,6 +14,9 @@ is an ispell-compatible spellchecker.
 \fB\-d \fIDICTIONARY\fR
 use the given dictionary
 .TP
+\fB\-p \fIWORDLIST\fR
+use the given personal wordlist
+.TP
 .B "\-a"
 list suggestions in ispell pipe mode format
 .TP
diff --git a/src/enchant.c b/src/enchant.c
index 06f72ce..2e520c0 100644
--- a/src/enchant.c
+++ b/src/enchant.c
@@ -1,7 +1,7 @@
 /* enchant
  * Copyright (C) 2003 Dom Lachowicz
  *               2007 Hannu Väisänen
- *               2016-2020 Reuben Thomas
+ *               2016-2021 Reuben Thomas
  *
  * This library is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
@@ -47,6 +47,7 @@
 #endif
 
 #include "enchant.h"
+#include "pwl.h"
 #include "enchant-provider.h"
 
 
@@ -75,6 +76,7 @@ print_help (const char * prog)
 	fprintf (stderr,
 		 "Usage: %s -a|-l|-h|-v [-L] [-d DICTIONARY] [FILE]\n\
   -d DICTIONARY  use the given dictionary\n\
+  -p FILE        use the given personal word list\n\
   -a             list suggestions in ispell pipe mode format\n\
   -l             list only the misspellings\n\
   -L             display line numbers\n\
@@ -132,10 +134,20 @@ print_utf (const char * str)
 	}
 }
 
+static int
+check_word (EnchantDict * dict, EnchantPWL * pwl, GString * word)
+{
+	int ok = word->len <= MIN_WORD_LENGTH ||
+		enchant_dict_check (dict, word->str, word->len) == 0;
+	if (!ok && pwl)
+		ok = enchant_pwl_check (pwl, word->str, word->len) == 0;
+	return ok;
+}
+
 static void
-do_mode_a (EnchantDict * dict, GString * word, size_t start_pos, size_t lineCount, gboolean terse_mode)
+do_mode_a (EnchantDict * dict, EnchantPWL * pwl, GString * word, size_t start_pos, size_t lineCount, gboolean terse_mode)
 {
-	if (word->len <= MIN_WORD_LENGTH || enchant_dict_check (dict, word->str, word->len) == 0) {
+	if (check_word (dict, pwl, word)) {
 		if (!terse_mode) {
 			if (lineCount)
 				printf ("* %u\n", (unsigned int)lineCount);
@@ -145,6 +157,8 @@ do_mode_a (EnchantDict * dict, GString * word, size_t start_pos, size_t lineCoun
 	} else {
 		size_t n_suggs;
 		char ** suggs = enchant_dict_suggest (dict, word->str, word->len, &n_suggs);
+		if (pwl)
+			suggs = enchant_pwl_suggest (pwl, word->str, word->len, suggs, &n_suggs);
 		if (!n_suggs || !suggs) {
 			printf ("# ");
 			if (lineCount)
@@ -174,9 +188,9 @@ do_mode_a (EnchantDict * dict, GString * word, size_t start_pos, size_t lineCoun
 }
 
 static void
-do_mode_l (EnchantDict * dict, GString * word, size_t lineCount)
+do_mode_l (EnchantDict * dict, EnchantPWL * pwl, GString * word, size_t lineCount)
 {
-	if (enchant_dict_check (dict, word->str, word->len) != 0) {
+	if (!check_word (dict, pwl, word)) {
 		if (lineCount)
 			printf ("%u ", (unsigned int)lineCount);
 		print_utf (word->str);
@@ -238,7 +252,7 @@ tokenize_line (EnchantDict * dict, GString * line)
 }
 
 static int
-parse_file (FILE * in, IspellMode_t mode, gboolean countLines, gchar *dictionary)
+parse_file (FILE * in, IspellMode_t mode, gboolean countLines, gchar *dictionary, EnchantPWL *pwl)
 {
 	EnchantBroker * broker;
 	EnchantDict * dict;
@@ -301,7 +315,10 @@ parse_file (FILE * in, IspellMode_t mode, gboolean countLines, gchar *dictionary
 				case '*': /* Insert in personal word list */
 					if (str->len == 1)
 						goto empty_word;
-					enchant_dict_add(dict, str->str + 1, -1);
+					if (pwl)
+						enchant_pwl_add (pwl, str->str + 1, -1);
+					else
+						enchant_dict_add (dict, str->str + 1, -1);
 					break;
 				case '@': /* Accept for this session */
 					if (str->len == 1)
@@ -311,7 +328,10 @@ parse_file (FILE * in, IspellMode_t mode, gboolean countLines, gchar *dictionary
 				case '/': /* Remove from personal word list */
 					if (str->len == 1)
 						goto empty_word;
-					enchant_dict_remove (dict, str->str + 1, -1);
+					if (pwl)
+						enchant_pwl_remove (pwl, str->str + 1, -1);
+					else
+						enchant_dict_remove (dict, str->str + 1, -1);
 					break;
 				case '_': /* Remove from this session */
 					if (str->len == 1)
@@ -375,9 +395,9 @@ parse_file (FILE * in, IspellMode_t mode, gboolean countLines, gchar *dictionary
 					tokens = tokens->next;
 
 					if (mode == MODE_A)
-						do_mode_a (dict, word, pos, lineCount, terse_mode);
+						do_mode_a (dict, pwl, word, pos, lineCount, terse_mode);
 					else if (mode == MODE_L)
-						do_mode_l (dict, word, lineCount);
+						do_mode_l (dict, pwl, word, lineCount);
 
 					g_string_free(word, TRUE);
 				}
@@ -409,6 +429,7 @@ int main (int argc, char ** argv)
 	FILE * fp = stdin;
 	gboolean countLines = FALSE;
 	gchar *dictionary = NULL;  /* -d dictionary */
+	gchar *perslist = NULL ; /* -p personal_word_list */
 
 	/* Initialize system locale */
 	setlocale(LC_ALL, "");
@@ -422,11 +443,14 @@ int main (int argc, char ** argv)
 #endif
 
 	int optchar;
-	while ((optchar = getopt (argc, argv, ":d:alvLmB")) != -1) {
+	while ((optchar = getopt (argc, argv, ":d:p:alvLmB")) != -1) {
 		switch (optchar) {
 		case 'd':
 			dictionary = optarg;  /* Emacs calls ispell with '-d dictionary'. */
 			break;
+		case 'p':
+			perslist = optarg;
+			break;
 		/* The first mode specified takes precedence. */
 		case 'a':
 			if (mode == MODE_NONE)
@@ -478,7 +502,17 @@ int main (int argc, char ** argv)
 			exit (1);
 		}
 	}
-	rval = parse_file (fp, mode, countLines, dictionary);
+	EnchantPWL *pwl = NULL;
+	if (perslist) {
+		pwl = enchant_pwl_init_with_file (perslist);
+		if (!pwl) {
+			fprintf (stderr, "Error: Could not read the personal word list \"%s\"", perslist);
+			exit (1);
+		}
+	}
+	rval = parse_file (fp, mode, countLines, dictionary, pwl);
+	if (pwl)
+		enchant_pwl_free (pwl);
 	if (file)
 		fclose (fp);
 	
diff --git a/src/pwl.c b/src/pwl.c
index 4f118a1..db645d5 100644
--- a/src/pwl.c
+++ b/src/pwl.c
@@ -313,8 +313,11 @@ static void enchant_pwl_remove_from_trie(EnchantPWL *pwl,
 }
 
 void enchant_pwl_add(EnchantPWL *pwl,
-			 const char *const word, size_t len)
+			 const char *const word, ssize_t len)
 {
+	if (len < 0)
+		len = strlen (word);
+
 	enchant_pwl_refresh_from_file(pwl);
 
 	enchant_pwl_add_to_trie(pwl, word, len);
@@ -342,7 +345,7 @@ void enchant_pwl_add(EnchantPWL *pwl,
 							putc ('\n', f);
 					}
 
-				if (fwrite (word, sizeof(char), len, f) == len)
+				if (fwrite (word, sizeof(char), len, f) == (size_t)len)
 					{
 						putc ('\n', f);
 					}
@@ -353,8 +356,11 @@ void enchant_pwl_add(EnchantPWL *pwl,
 }
 
 void enchant_pwl_remove(EnchantPWL *pwl,
-			 const char *const word, size_t len)
+			 const char *const word, ssize_t len)
 {
+	if (len < 0)
+		len = strlen (word);
+
 	if(enchant_pwl_check(pwl, word, len) == 1)
 		return;
 
@@ -525,8 +531,11 @@ static gchar* enchant_utf8_strtitle(const gchar*str, gssize len)
 	return result;
 }
 
-int enchant_pwl_check(EnchantPWL *pwl, const char *const word, size_t len)
+int enchant_pwl_check(EnchantPWL *pwl, const char *const word, ssize_t len)
 {
+	if (len < 0)
+		len = strlen (word);
+
 	enchant_pwl_refresh_from_file(pwl);
 
 	int exists = enchant_pwl_contains(pwl, word, len);
@@ -609,8 +618,11 @@ static int best_distance(char** suggs, const char *const word, size_t len)
 /* gives the best set of suggestions from pwl that are at least as good as the 
  * given suggs (if suggs == NULL just best from pwl) */
 char** enchant_pwl_suggest(EnchantPWL *pwl, const char *const word,
-			   size_t len, char** suggs, size_t* out_n_suggs)
+			   ssize_t len, char** suggs, size_t* out_n_suggs)
 {
+	if (len < 0)
+		len = strlen (word);
+
 	int max_dist = suggs ? best_distance(suggs, word, len) : ENCHANT_PWL_MAX_ERRORS;
 	max_dist = MIN (max_dist, ENCHANT_PWL_MAX_ERRORS);
 
diff --git a/src/pwl.h b/src/pwl.h
index f713b29..c945e3c 100644
--- a/src/pwl.h
+++ b/src/pwl.h
@@ -42,12 +42,12 @@ typedef struct str_enchant_pwl EnchantPWL;
 EnchantPWL* enchant_pwl_init(void);
 EnchantPWL* enchant_pwl_init_with_file(const char * file);
 
-void enchant_pwl_add(EnchantPWL * me, const char *const word, size_t len);
-void enchant_pwl_remove(EnchantPWL * me, const char *const word, size_t len);
-int enchant_pwl_check(EnchantPWL * me,const char *const word, size_t len);
+void enchant_pwl_add(EnchantPWL * me, const char *const word, ssize_t len);
+void enchant_pwl_remove(EnchantPWL * me, const char *const word, ssize_t len);
+int enchant_pwl_check(EnchantPWL * me, const char *const word, ssize_t len);
 /*gives the best set of suggestions from pwl that are at least as good as the given suggs*/
 char** enchant_pwl_suggest(EnchantPWL *me, const char *const word,
-			   size_t len, char ** suggs, size_t* out_n_suggs);
+			   ssize_t len, char ** suggs, size_t* out_n_suggs);
 void enchant_pwl_free(EnchantPWL* me);
 
 #ifdef __cplusplus