char array cleanup

rlpowell · rlpowell · commit 398bbc0edf8a · 2020-11-29T21:10:46.000-08:00
- made many char[] arrays larger; see #13 for why. This was done without any attempt to understand the code, so please report any problems. - cleaned up some gcc warnings
diff --git a/NEWS b/NEWS
@@ -1,3 +1,12 @@
+New in version 0.42
+===================
+
+- made many char[] arrays larger; see
+  https://github.com/lojban/jbofihe/issues/13 for why.  This was
+  done without any attempt to understand the code, so please report
+  any problems.
+- cleaned up some gcc warnings
+
 New in version 0.41
 ===================
 
diff --git a/build-in-container.sh b/build-in-container.sh
@@ -11,6 +11,6 @@ make all
 DESTDIR=build/ make install
 rm -rf build/usr/lib/.build-id
 cp find_lojban.pl build/usr/local/bin/find_lojban
-rm -f jbofihe-0.41-1.x86_64.rpm
-fpm -s dir -t rpm -n jbofihe -v 0.41 --rpm-rpmbuild-define "_build_id_links none" -C build/ usr/local
-rpm -qpl jbofihe-0.41-1.x86_64.rpm | sort
+rm -f jbofihe-0.42-1.x86_64.rpm
+fpm -s dir -t rpm -n jbofihe -v 0.42 --rpm-rpmbuild-define "_build_id_links none" -C build/ usr/local
+rpm -qpl jbofihe-0.42-1.x86_64.rpm | sort
diff --git a/canonluj.c b/canonluj.c
@@ -120,7 +120,7 @@ lookup_component(char *x)
 static void
 strip_leading_rafsi(char **y, int len, int hy, int first, char *result)
 {
-  char buf[5], *comp;
+  char buf[65536], *comp;
   char *x = *y;
   int i;
   for (i=0; i<len; i++) buf[i] = x[i];
@@ -138,7 +138,7 @@ strip_leading_rafsi(char **y, int len, int hy, int first, char *result)
 static char *
 split_lujvo (char *x)
 {
-  static char result[2048];
+  static char result[65536];
   int first=1; /* At start, so we can handle r and n hyphens */
   
   result[0] = 0;
@@ -279,7 +279,7 @@ canon_lujvo (char *x)
 #ifdef TEST
 
 int main (int argc, char **argv) {
-  char line[128], *result;
+  char line[65536], *result;
   while (fgets(line, sizeof(line), stdin)) {
     line[strlen(line)-1] = 0; /* strip NL */
     result = canon_lujvo(line);
diff --git a/cm_gather.c b/cm_gather.c
@@ -105,8 +105,8 @@ static void
 try_cmavo_fragment(node *p)
 {
   char buf_in[32];
-  char buf[2048];
-  char buf_out[256];
+  char buf[65536];
+  char buf_out[65536];
 
   char *q, *r, *t;
   buf[0] = 0;
@@ -146,10 +146,10 @@ translate_selmao(char *x) {
 static void
 try_cmavo_fragment(char *lojban, char **transbuf, char* (*transfn)(char *))
 {
-  char buf[256], buf2[256];
+  char buf[65536], buf2[65536];
   char *pbuf;
   char *t, *e;
-  char bufout[1024];
+  char bufout[65536];
   int len;
 
   bufout[0] = 0;
diff --git a/cm_translate.c b/cm_translate.c
@@ -22,7 +22,7 @@
 char *
 translate(char *word)
 {
-  static char buf[1024];
+  static char buf[65536];
   char *res;
 
   res = dict_lookup(word);
@@ -63,7 +63,7 @@ static char *
 translate_comp(char *w, int *conversion)
 {
   char buf[64], buf2[4];
-  static char result[256];
+  static char result[65536];
   char *trans;
   strcpy(buf, w);
   if (*conversion) {
@@ -109,7 +109,7 @@ translate_comp(char *w, int *conversion)
 static char *
 translate_lujvo(char *word)
 {
-  static char result[4096];
+  static char result[65536];
   char buf[64];
   char *w, *trans;
   int initial;
@@ -265,7 +265,7 @@ is_consonant_not_r(char c)
 char *
 translate_unknown(char *w)
 {
-  static char buf[2048];
+  static char buf[65536];
   int len, i;
   int hyph;
   char *p, *q;
diff --git a/conversion.c b/conversion.c
@@ -15,6 +15,7 @@
 */
 
 #include <assert.h>
+#include <string.h>
 #include "nodes.h"
 #include "functions.h"
 #include "rpc2x_act.tab.h"
diff --git a/functions.c b/functions.c
@@ -675,7 +675,7 @@ is_simple_nonterm(TreeNode *x)
 char *
 make_uppercase(char *s)
 {
-  static char buffer[1024];
+  static char buffer[65536];
   char *p;
   strcpy(buffer, s);
   p = buffer;
diff --git a/htmlout.c b/htmlout.c
@@ -224,7 +224,7 @@ write_close_bracket(BracketType type, int subscript)
 static char *
 make_htmlsafe(char *s)
 {
-  static char buf[2048];
+  static char buf[65536];
   char *p, *q;
   p = s;
   q = buf;
diff --git a/latex.c b/latex.c
@@ -260,7 +260,7 @@ latex_write_close_bracket(BracketType type, int subscript)
 char *
 make_texsafe(char *s)
 {
-  static char buf[2048];
+  static char buf[65536];
   char *p, *q;
   p = s;
   q = buf;
diff --git a/latexblk.c b/latexblk.c
@@ -25,9 +25,9 @@ typedef enum {
 
 static States state;
 
-static char tag_text[4096];
-static char loj_text[4096];
-static char eng_text[4096];
+static char tag_text[65536];
+static char loj_text[65536];
+static char eng_text[65536];
 
 /*++++++++++++++++++++++++++++++++++++++
   
@@ -411,7 +411,7 @@ start_tag(void)
 static void
 write_tag_text(char *brivla, char *place, char *trans, int brac)
 {
-  char buffer[1024];
+  char buffer[65536];
 
   if (brac) {
     sprintf(buffer, "%s%s (%s)", brivla, place, make_texsafe(trans));
@@ -426,7 +426,7 @@ write_tag_text(char *brivla, char *place, char *trans, int brac)
 
 static void write_partial_tag_text(char *t)/*{{{*/
 {
-  char buffer[1024];
+  char buffer[65536];
 
   sprintf(buffer, "%s", t);
   if (tag_text[1]) { /* '[' is pre-inserted by start_tags */
diff --git a/lex1.c b/lex1.c
@@ -185,7 +185,7 @@ process_cmene(char *buf, int start_line, int start_column,
   TreeNode *tok;
 
   if (is_bad)  {
-    char prefix[1024], labuf[8];
+    char prefix[65536], labuf[8];
     char *p, *q;
     int len1, len2;
     for (p=buf, q=prefix, len1=0; p!=ladoi; len1++) *q++ = *p++;
@@ -216,7 +216,7 @@ add_preceding_cmavo(char **pws, char **pwe, int start_line, int *column)
   char **pw;
   char *p, *q;
   int incr;
-  char buf2[1024];
+  char buf2[65536];
   for (pw=pws; pw<pwe; pw++) {
     incr = 0;
     for (p=*pw, q=buf2; p<*(pw+1);) {
@@ -241,9 +241,9 @@ add_preceding_cmavo(char **pws, char **pwe, int start_line, int *column)
 static int
 process_word(char *buf, int start_line, int start_column)
 {
-  char buf2[1024];
+  char buf2[65536];
   TreeNode *tok;
-  char *word_starts[1024];
+  char *word_starts[65536];
   char **pws, **pwe;
   MorfType morf_type;
   struct morf_xtra mx;
@@ -379,7 +379,7 @@ process_word(char *buf, int start_line, int start_column)
 void
 parse_file(FILE *f)
 {
-  char buffer[1024]; /* buffer for words */
+  char buffer[65536]; /* buffer for words */
   char *p;
   int c;
   int line;
diff --git a/lujvofns.c b/lujvofns.c
@@ -243,7 +243,7 @@ int
 is_valid_lujvo(char *t)
 {
   int len = strlen(t);
-  char buf[1024];
+  char buf[65536];
 
   /* Flags to check CVV + hyphen stuff. */
   int start_cvv;
diff --git a/morf.c b/morf.c
@@ -62,6 +62,7 @@ enum processed_category {/*{{{*/
 
 /* Include file for checking vowel pairs/clusters within the input stream. */
 #include "morfvlex.c"
+#include <stdlib.h>
 
 static unsigned char s2l[32] = /*{{{*/
 /* Map N->1, R->2, other C->3, else ->0.  Used to trim down the last-but-one
@@ -209,7 +210,7 @@ MorfType morf_scan(char *s, char ***buf_end, struct morf_xtra *arg_xtra)/*{{{*/
   unsigned int vsm = 0111; /* 3 consonants as starting state */
   char *p, c;
   char **start, **pstart;
-  char *cstart[256], **pcstart; /* Used for holding start of Cy structures */
+  char *cstart[65536], **pcstart; /* Used for holding start of Cy structures */
   int ent, tent, tok, set, inhibit;
   int state, next_state;
   int inhibited = 0, initial = 1;
@@ -686,8 +687,8 @@ MorfType morf_scan(char *s, char ***buf_end, struct morf_xtra *arg_xtra)/*{{{*/
 
 #ifdef TEST_MORF
 int main (int argc, char **argv) {/*{{{*/
-  char buffer[128];
-  char *start[256], **pstart;
+  char buffer[65536];
+  char *start[65536], **pstart;
   char *word = NULL;
   while (++argv, --argc) {
     if (!strncmp(*argv, "-v", 2)) {
diff --git a/output.c b/output.c
@@ -186,7 +186,7 @@ static void translate_bai (TreeNode *x, char *eng)/*{{{*/
 {
   XBaiConversion *baiconv;
   XTenseCtx *xtc;
-  char buffer[1024], *trans;
+  char buffer[65536], *trans;
 
   eng[0] = 0;
   buffer[0] = 0;
@@ -336,7 +336,7 @@ static void get_cmavo_text_inside_node_internal(TreeNode *x, char *sofar)/*{{{*/
 /*}}}*/
 static char * get_cmavo_text_inside_node(TreeNode *x)/*{{{*/
 {
-  static char buffer[4096];
+  static char buffer[65536];
   buffer[0] = 0;
   get_cmavo_text_inside_node_internal(x, buffer);
   return buffer;
@@ -358,7 +358,7 @@ static void translate_indicator (TreeNode *x, char *loj, char *eng)/*{{{*/
 {
   char *trans;
   int negated;
-  char buffer[256];
+  char buffer[65536];
   XCaiIndicator *xci;
 
   if (prop_neg_indicator(x, NO)) {
@@ -453,7 +453,7 @@ static void translate_indicator (TreeNode *x, char *loj, char *eng)/*{{{*/
 /*}}}*/
 static char * translate_tense_in_context(char *text, enum tense_contexts ctx)/*{{{*/
 {
-  char buffer[128];
+  char buffer[65536];
   char *trans;
 
   strcpy(buffer, text);
@@ -504,7 +504,7 @@ static char * translate_tense_in_context(char *text, enum tense_contexts ctx)/*{
 static void translate_tense (TreeNode *x, char *eng)/*{{{*/
 {
   XTenseCtx *ctx;
-  char buffer[128], *trans;
+  char buffer[65536], *trans;
 
   eng[0] = 0;
   ctx = prop_tense_ctx(x, NO);
@@ -553,7 +553,7 @@ static void translate_time_offset (TreeNode *x, char *loj, char *eng)/*{{{*/
     if (trans) {
       strcpy(eng, trans);
     } else {
-      char engbuf[256];
+      char engbuf[65536];
       engbuf[0] = 0;
       trans = translate(cmavo_table[pu->data.cmavo.code].cmavo);
       strcat(engbuf, trans ? trans : "?");
@@ -580,7 +580,7 @@ static void translate_time_offset (TreeNode *x, char *loj, char *eng)/*{{{*/
 static void translate_jai (TreeNode *x, char *eng)/*{{{*/
 {
   XGlosstype *xgt;
-  char buffer[128], *trans;
+  char buffer[65536], *trans;
 
   strcpy(buffer, cmavo_table[x->data.cmavo.code].cmavo);
   eng[0] = 0;
@@ -614,7 +614,7 @@ static void translate_jai (TreeNode *x, char *eng)/*{{{*/
 /*}}}*/
 static char * translate_convertible(char *loj, TreeNode *basis)/*{{{*/
 {
-  char buffer[1024];
+  char buffer[65536];
   char *trans;
   XConversion *conversion;
   XGlosstype *gt;
@@ -775,7 +775,7 @@ static void get_lojban_word_and_translation (TreeNode *x, char *loj, char *eng)/
       
     case N_ZO:
       {
-        char buffer[1024];
+        char buffer[65536];
         sprintf(loj, "zo %s", x->data.zo.text);
         attempt_translation(x->data.zo.text, buffer);
         sprintf(eng, "\"%s\"", buffer);
@@ -869,7 +869,7 @@ static void output_term(TreeNode *x, WhatToShow what)/*{{{*/
             break;/*}}}*/
           case TTT_JAITAG:/*{{{*/
             {
-              char transbuf[1024];
+              char transbuf[65536];
               (drv->start_tag)();
               (drv->write_partial_tag_text)("jai+<tag>1: (");
               output_internal(tag->jaitag.tag, SHOW_TAG_TRANS);
@@ -900,7 +900,7 @@ static void output_term(TreeNode *x, WhatToShow what)/*{{{*/
 /*}}}*/
           case TTT_ME:/*{{{*/
             {
-              char *trans, transbuf[1024];
+              char *trans, transbuf[65536];
               (drv->start_tag)();
               trans = adv_translate("me", tag->pos, TCX_TAG);
               sprintf(transbuf, trans, tag->me.sumti->data.nonterm.number);
@@ -910,7 +910,7 @@ static void output_term(TreeNode *x, WhatToShow what)/*{{{*/
           break;/*}}}*/
           case TTT_NUMBERMOI:/*{{{*/
             {
-              char *trans, lojbuf[128], transbuf[1024];
+              char *trans, lojbuf[65536], transbuf[65536];
               int code;
               int norl_code;
               char *cmavo;
@@ -984,7 +984,7 @@ static void output_term(TreeNode *x, WhatToShow what)/*{{{*/
 /*}}}*/
 static void output_simple_time_offset(TreeNode *x, WhatToShow what)/*{{{*/
 {
-  char loj[1024], eng[1024];
+  char loj[65536], eng[65536];
   int i, n;
 
   loj[0] = 0;
@@ -1187,8 +1187,8 @@ static void output_fore_or_afterthought(TreeNode *x, WhatToShow what)/*{{{*/
 static void output_clustered(TreeNode *x, WhatToShow what)/*{{{*/
 {
   char *cluster, *trans;
-  char localtrans[256];
-  char lojbuf[1024], lojbuf2[1024];
+  char localtrans[65536];
+  char lojbuf[65536], lojbuf2[65536];
   int i, n;
   TreeNode *c;
   struct nonterm *nt;
@@ -1277,7 +1277,7 @@ static void output_clustered(TreeNode *x, WhatToShow what)/*{{{*/
 /*}}}*/
 static void output_internal(TreeNode *x, WhatToShow what)/*{{{*/
 {
-  char loj[1024], eng[1024];
+  char loj[65536], eng[65536];
   int i, n;
 
   if (x->type == N_NONTERM) {/*{{{*/
@@ -1440,7 +1440,7 @@ static void output_internal(TreeNode *x, WhatToShow what)/*{{{*/
 /*}}}*/
   } else {/*{{{*/
     /* Terminal token */
-    char lojbuf[1024];
+    char lojbuf[65536];
 
     lojbuf[0] = 0;
 
diff --git a/relative.c b/relative.c
diff --git a/smujajgau.c b/smujajgau.c
diff --git a/textblk.c b/textblk.c
diff --git a/translate.c b/translate.c
diff --git a/tree.c b/tree.c
diff --git a/version.txt b/version.txt

Original file line number	Diff line number	Diff line change
`@@ -675,7 +675,7 @@ is_simple_nonterm(TreeNode *x)`
`675`	`675`	`char *`
`676`	`676`	`make_uppercase(char *s)`
`677`	`677`	`{`
`678`		`- static char buffer[1024];`
	`678`	`+ static char buffer[65536];`
`679`	`679`	`char *p;`
`680`	`680`	`strcpy(buffer, s);`
`681`	`681`	`p = buffer;`
Original file line number	Diff line number	Diff line change
`@@ -224,7 +224,7 @@ write_close_bracket(BracketType type, int subscript)`
`224`	`224`	`static char *`
`225`	`225`	`make_htmlsafe(char *s)`
`226`	`226`	`{`
`227`		`- static char buf[2048];`
	`227`	`+ static char buf[65536];`
`228`	`228`	`char p, q;`
`229`	`229`	`p = s;`
`230`	`230`	`q = buf;`
Original file line number	Diff line number	Diff line change
`@@ -260,7 +260,7 @@ latex_write_close_bracket(BracketType type, int subscript)`
`260`	`260`	`char *`
`261`	`261`	`make_texsafe(char *s)`
`262`	`262`	`{`
`263`		`- static char buf[2048];`
	`263`	`+ static char buf[65536];`
`264`	`264`	`char p, q;`
`265`	`265`	`p = s;`
`266`	`266`	`q = buf;`