Check syntax of obsolete entries of PO files, not only in msgmerge.

4 files changed, 167 insertions, 82 deletions

diff --git a/src/ChangeLog b/src/ChangeLog
index cf4062b..ec4aed3 100644
--- a/src/ChangeLog
+++ b/src/ChangeLog
@@ -1,3 +1,28 @@
+2001-03-04  Bruno Haible  <haible@clisp.cons.org>
+
+	Check syntax of obsolete entries of PO files, not only in msgmerge.
+	* po-lex.h (pass_obsolete_entries): New declaration.
+	(msgstr_def): Remove pos field.
+	* po-lex.c (po_lex_obsolete): New variable.
+	(pass_obsolete_entries): Make non-static.
+	(lex_open): Initialize po_lex_obsolete.
+	(lex_close): Reset po_lex_obsolete.
+	(po_gram_lex): Don't look at pass_obsolete_entries. Instead, set
+	po_lex_obsolete to 1 when "#~" is seen. Reset po_lex_obsolete when
+	a newline is seen. Before returning, fill the 'pos' and 'obsolete'
+	fields of any token.
+	* po-gram.gen.y (check_obsolete): New macro.
+	(union): Add a 'pos' and 'obsolete' field for any symbol type.
+	(NAME): Assign to type <string>.
+	(DOMAIN, MSGID, MSGID_PLURAL, MSGSTR, '[', ']'): Assign to type <pos>.
+	(msgid, msgstr): Remove.
+	(message, msgid_pluralform, pluralform_list, pluralform, string_list):
+	Signal an error if the 'obsolete' field is not the same across the
+	entire token sequence of the rule.
+	(message): Deal with pass_obsolete_entries here.
+	(msgid_pluralform, pluralform_list, pluralform, string_list): Fill the
+	'pos' and 'obsolete' fields of $$.
+
 2001-03-03  Bruno Haible  <haible@clisp.cons.org>
 
 	Fix parsing of strings in CJK encodings.
diff --git a/src/po-gram-gen.y b/src/po-gram-gen.y
index ddd776d..1b576aa 100644
--- a/src/po-gram-gen.y
+++ b/src/po-gram-gen.y
@@ -81,6 +81,11 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
 #define yycheck  po_gram_yycheck
 
 static long plural_counter;
+
+#define check_obsolete(value1,value2) \
+  if ((value1).obsolete != (value2).obsolete) \
+    po_gram_error_at_line (&(value2).pos, _("inconsistent use of #~"));
+
 %}
 
 %token	COMMENT
@@ -96,15 +101,15 @@ static long plural_counter;
 
 %union
 {
-  char *string;
-  long number;
-  lex_pos_ty pos;
-  struct msgstr_def rhs;
+  struct { char *string; lex_pos_ty pos; int obsolete; } string;
+  struct { long number; lex_pos_ty pos; int obsolete; } number;
+  struct { lex_pos_ty pos; int obsolete; } pos;
+  struct { struct msgstr_def rhs; lex_pos_ty pos; int obsolete; } rhs;
 }
 
-%type <string> STRING COMMENT string_list msgid_pluralform
+%type <string> STRING COMMENT NAME string_list msgid_pluralform
 %type <number> NUMBER
-%type <pos> msgid msgstr
+%type <pos> DOMAIN MSGID MSGID_PLURAL MSGSTR '[' ']'
 %type <rhs> pluralform pluralform_list
 
 %right MSGSTR
@@ -122,45 +127,72 @@ msgfmt
 domain
 	: DOMAIN STRING
 		{
-		   po_callback_domain ($2);
+		   po_callback_domain ($2.string);
 		}
 	;
 
 message
-	: msgid string_list msgstr string_list
+	: MSGID string_list MSGSTR string_list
 		{
-		  po_callback_message ($2, &$1, NULL,
-				       $4, strlen ($4) + 1, &$3);
+		  check_obsolete ($1, $2);
+		  check_obsolete ($1, $3);
+		  check_obsolete ($1, $4);
+		  if (!$1.obsolete || pass_obsolete_entries)
+		    po_callback_message ($2.string, &$1.pos, NULL,
+					 $4.string, strlen ($4.string) + 1, &$3.pos);
+		  else
+		    {
+		      free ($2.string);
+		      free ($4.string);
+		    }
 		}
-	| msgid string_list msgid_pluralform pluralform_list
+	| MSGID string_list msgid_pluralform pluralform_list
 		{
-		  po_callback_message ($2, &$1, $3,
-				       $4.msgstr, $4.msgstr_len, &$4.pos);
+		  check_obsolete ($1, $2);
+		  check_obsolete ($1, $3);
+		  check_obsolete ($1, $4);
+		  if (!$1.obsolete || pass_obsolete_entries)
+		    po_callback_message ($2.string, &$1.pos, $3.string,
+					 $4.rhs.msgstr, $4.rhs.msgstr_len, &$4.pos);
+		  else
+		    {
+		      free ($2.string);
+		      free ($3.string);
+		      free ($4.rhs.msgstr);
+		    }
 		}
-	| msgid string_list msgid_pluralform
+	| MSGID string_list msgid_pluralform
 		{
-		  po_gram_error_at_line (&$1, _("missing `msgstr[]' section"));
-		  free ($2);
-		  free ($3);
+		  check_obsolete ($1, $2);
+		  check_obsolete ($1, $3);
+		  po_gram_error_at_line (&$1.pos, _("missing `msgstr[]' section"));
+		  free ($2.string);
+		  free ($3.string);
 		}
-	| msgid string_list pluralform_list
+	| MSGID string_list pluralform_list
 		{
-		  po_gram_error_at_line (&$1, _("missing `msgid_plural' section"));
-		  free ($2);
-		  free ($3.msgstr);
+		  check_obsolete ($1, $2);
+		  check_obsolete ($1, $3);
+		  po_gram_error_at_line (&$1.pos, _("missing `msgid_plural' section"));
+		  free ($2.string);
+		  free ($3.rhs.msgstr);
 		}
-	| msgid string_list
+	| MSGID string_list
 		{
-		  po_gram_error_at_line (&$1, _("missing `msgstr' section"));
-		  free ($2);
+		  check_obsolete ($1, $2);
+		  po_gram_error_at_line (&$1.pos, _("missing `msgstr' section"));
+		  free ($2.string);
 		}
 	;
 
 msgid_pluralform
 	: MSGID_PLURAL string_list
 		{
+		  check_obsolete ($1, $2);
 		  plural_counter = 0;
-		  $$ = $2;
+		  $$.string = $2.string;
+		  $$.pos = $1.pos;
+		  $$.obsolete = $1.obsolete;
 		}
 	;
 
@@ -171,44 +203,37 @@ pluralform_list
 		}
 	| pluralform_list pluralform
 		{
-		  $$.msgstr = (char *) xmalloc ($1.msgstr_len + $2.msgstr_len);
-		  memcpy ($$.msgstr, $1.msgstr, $1.msgstr_len);
-		  memcpy ($$.msgstr + $1.msgstr_len, $2.msgstr, $2.msgstr_len);
-		  $$.msgstr_len = $1.msgstr_len + $2.msgstr_len;
+		  check_obsolete ($1, $2);
+		  $$.rhs.msgstr = (char *) xmalloc ($1.rhs.msgstr_len + $2.rhs.msgstr_len);
+		  memcpy ($$.rhs.msgstr, $1.rhs.msgstr, $1.rhs.msgstr_len);
+		  memcpy ($$.rhs.msgstr + $1.rhs.msgstr_len, $2.rhs.msgstr, $2.rhs.msgstr_len);
+		  $$.rhs.msgstr_len = $1.rhs.msgstr_len + $2.rhs.msgstr_len;
+		  free ($1.rhs.msgstr);
+		  free ($2.rhs.msgstr);
 		  $$.pos = $1.pos;
-		  free ($1.msgstr);
-		  free ($2.msgstr);
+		  $$.obsolete = $1.obsolete;
 		}
 	;
 
 pluralform
-	: msgstr '[' NUMBER ']' string_list
+	: MSGSTR '[' NUMBER ']' string_list
 		{
-		  if ($3 != plural_counter)
+		  check_obsolete ($1, $2);
+		  check_obsolete ($1, $3);
+		  check_obsolete ($1, $4);
+		  check_obsolete ($1, $5);
+		  if ($3.number != plural_counter)
 		    {
 		      if (plural_counter == 0)
-			po_gram_error_at_line (&$1, _("first plural form has nonzero index"));
+			po_gram_error_at_line (&$1.pos, _("first plural form has nonzero index"));
 		      else
-			po_gram_error_at_line (&$1, _("plural form has wrong index"));
+			po_gram_error_at_line (&$1.pos, _("plural form has wrong index"));
 		    }
 		  plural_counter++;
-		  $$.msgstr = $5;
-		  $$.msgstr_len = strlen ($5) + 1;
-		  $$.pos = $1;
-		}
-	;
-
-msgid
-	: MSGID
-		{
-		  $$ = gram_pos;
-		}
-	;
-
-msgstr
-	: MSGSTR
-		{
-		  $$ = gram_pos;
+		  $$.rhs.msgstr = $5.string;
+		  $$.rhs.msgstr_len = strlen ($5.string) + 1;
+		  $$.pos = $1.pos;
+		  $$.obsolete = $1.obsolete;
 		}
 	;
 
@@ -222,18 +247,21 @@ string_list
 		  size_t len1;
 		  size_t len2;
 
-		  len1 = strlen ($1);
-		  len2 = strlen ($2);
-		  $$ = (char *) xmalloc (len1 + len2 + 1);
-		  stpcpy (stpcpy ($$, $1), $2);
-		  free ($1);
-		  free ($2);
+		  check_obsolete ($1, $2);
+		  len1 = strlen ($1.string);
+		  len2 = strlen ($2.string);
+		  $$.string = (char *) xmalloc (len1 + len2 + 1);
+		  stpcpy (stpcpy ($$.string, $1.string), $2.string);
+		  free ($1.string);
+		  free ($2.string);
+		  $$.pos = $1.pos;
+		  $$.obsolete = $1.obsolete;
 		}
 	;
 
 comment
 	: COMMENT
 		{
-		  po_callback_comment ($1);
+		  po_callback_comment ($1.string);
 		}
 	;
diff --git a/src/po-lex.c b/src/po-lex.c
index cf43fe7..509f802 100644
--- a/src/po-lex.c
+++ b/src/po-lex.c
@@ -59,12 +59,13 @@
 static FILE *fp;
 lex_pos_ty gram_pos;
 unsigned int gram_max_allowed_errors = 20;
+static int po_lex_obsolete;
 const char *po_lex_charset;
 #if HAVE_ICONV
 iconv_t po_lex_iconv;
 #endif
 static int pass_comments = 0;
-static int pass_obsolete_entries = 0;
+int pass_obsolete_entries = 0;
 
 
 /* Prototypes for local functions.  */
@@ -85,6 +86,7 @@ lex_open (fname)
 	   _("error while opening \"%s\" for reading"), fname);
 
   gram_pos.line_number = 1;
+  po_lex_obsolete = 0;
   po_lex_charset = NULL;
 #if HAVE_ICONV
   po_lex_iconv = (iconv_t)(-1);
@@ -105,6 +107,7 @@ lex_close ()
   gram_pos.file_name = 0;
   gram_pos.line_number = 0;
   error_message_count = 0;
+  po_lex_obsolete = 0;
   po_lex_charset = NULL;
 #if HAVE_ICONV
   if (po_lex_iconv != (iconv_t)(-1))
@@ -374,7 +377,7 @@ control_sequence ()
 
 
 /* Return the next token in the PO file.  The return codes are defined
-   in "po-gram-gen2.h".  Associated data is put in 'po_gram_lval.  */
+   in "po-gram-gen2.h".  Associated data is put in 'po_gram_lval'.  */
 int
 po_gram_lex ()
 {
@@ -392,27 +395,32 @@ po_gram_lex ()
 	  /* Yacc want this for end of file.  */
 	  return 0;
 
+	case '\n':
+	  po_lex_obsolete = 0;
+	  break;
+
 	case ' ':
 	case '\t':
-	case '\n':
 	case '\r':
 	case '\f':
 	case '\v':
 	  break;
 
 	case '#':
+	  c = lex_getc ();
+	  if (c == '~')
+	    /* A pseudo-comment beginning with #~ is found.  This is
+	       not a comment.  It is the format for obsolete entries.
+	       We simply discard the "#~" prefix.  The following
+	       characters are expected to be well formed.  */
+	    {
+	      po_lex_obsolete = 1;
+	      break;
+	    }
+
 	  /* Accumulate comments into a buffer.  If we have been asked
  	     to pass comments, generate a COMMENT token, otherwise
  	     discard it.  */
-	  c = lex_getc ();
-	  if (c == '~' && pass_obsolete_entries)
-	    /* A special comment beginning with #~ is found.  This
-	       is the format for obsolete entries and if we are
-	       asked to return them is entries not as comments be
-	       simply stop processing the comment here.  The
-	       following characters are expected to be well formed.  */
-	    break;
-
 	  if (pass_comments)
 	    {
 	      bufpos = 0;
@@ -431,15 +439,21 @@ po_gram_lex ()
 		}
 	      buf[bufpos] = 0;
 
-	      po_gram_lval.string = buf;
+	      po_gram_lval.string.string = buf;
+	      po_gram_lval.string.pos = gram_pos;
+	      po_gram_lval.string.obsolete = po_lex_obsolete;
+	      po_lex_obsolete = 0;
 	      return COMMENT;
 	    }
 	  else
-	    /* We do this in separate loop because collecting large
-	       comments while they get not passed to the upper layers
-	       is not very effective.  */
-	    while (c != EOF && c != '\n')
-	      c = lex_getc ();
+	    {
+	      /* We do this in separate loop because collecting large
+		 comments while they get not passed to the upper layers
+		 is not very effective.  */
+	      while (c != EOF && c != '\n')
+		c = lex_getc ();
+	      po_lex_obsolete = 0;
+	    }
 	  break;
 
 	case '"':
@@ -519,7 +533,9 @@ po_gram_lex ()
 	    buf[bufpos] = 0;
 
 	    /* FIXME: Treatment of embedded \000 chars is incorrect.  */
-	    po_gram_lval.string = xstrdup (buf);
+	    po_gram_lval.string.string = xstrdup (buf);
+	    po_gram_lval.string.pos = gram_pos;
+	    po_gram_lval.string.obsolete = po_lex_obsolete;
 	    return STRING;
 	  }
 
@@ -576,12 +592,20 @@ po_gram_lex ()
 
 	  c = keyword_p (buf);
 	  if (c == NAME)
-	    po_gram_lval.string = xstrdup (buf);
+	    {
+	      po_gram_lval.string.string = xstrdup (buf);
+	      po_gram_lval.string.pos = gram_pos;
+	      po_gram_lval.string.obsolete = po_lex_obsolete;
+	    }
+	  else
+	    {
+	      po_gram_lval.pos.pos = gram_pos;
+	      po_gram_lval.pos.obsolete = po_lex_obsolete;
+	    }
 	  return c;
 
 	case '0': case '1': case '2': case '3': case '4':
 	case '5': case '6': case '7': case '8': case '9':
-	  /* I know, we don't need numbers, yet.  */
 	  bufpos = 0;
 	  for (;;)
 	    {
@@ -608,13 +632,19 @@ po_gram_lex ()
 
 	  buf[bufpos] = 0;
 
-	  po_gram_lval.number = atol (buf);
+	  po_gram_lval.number.number = atol (buf);
+	  po_gram_lval.number.pos = gram_pos;
+	  po_gram_lval.number.obsolete = po_lex_obsolete;
 	  return NUMBER;
 
 	case '[':
+	  po_gram_lval.pos.pos = gram_pos;
+	  po_gram_lval.pos.obsolete = po_lex_obsolete;
 	  return '[';
 
 	case ']':
+	  po_gram_lval.pos.pos = gram_pos;
+	  po_gram_lval.pos.obsolete = po_lex_obsolete;
 	  return ']';
 
 	default:
diff --git a/src/po-lex.h b/src/po-lex.h
index 4beb7bc..279382a 100644
--- a/src/po-lex.h
+++ b/src/po-lex.h
@@ -47,6 +47,9 @@ extern const char *po_lex_charset;
 extern iconv_t po_lex_iconv;
 #endif
 
+/* Nonzero if obsolete entries shall be considered as valid.  */
+extern int pass_obsolete_entries;
+
 
 /* Open the PO file FNAME and prepare its lexical analysis.  */
 extern void lex_open PARAMS ((const char *__fname));
@@ -138,7 +141,6 @@ struct msgstr_def
 {
   char *msgstr;
   size_t msgstr_len;
-  lex_pos_ty pos;
 };
 
 #endif