poe

чување d1019afbfcd2a3efc2dcf18a03d1d285d28f0d72
родитељ 49f04f3225ed8a22d58ace022f10af1c2e83cb28
Аутор: Страхиња Радић <sr@strahinja.org>
Датум:   Thu, 12 Feb 2026 18:56:09 +0100

po.c(load_file): Improve binary file and syntax check; (parse_po_line): Improve parsing

Diffstat:
M
po.c
 | 
270
+++++++++++++++++++++++++++++++++++++++++++++++++------------------------------
M
po.h
 | 
41
++++++++++++++++++++---------------------
M
poe.c
 | 
13
+++++++++++--

измењених датотека: 3, додавања: 200(+), брисања: 124(-)
diff --git a/po.c b/po.c
@@ -421,18 +421,19 @@ add_ref_comment(struct PoEntry* entry, const char* comment)
 int
 load_file(struct DrawState* dstate, long* lineno, long* col)
 {
-	struct PoEntry* newchunk = NULL;
-	struct PoEntry* entry	 = NULL;
-	struct PoEntry* current	 = NULL;
-	FILE* input		 = NULL;
-	char* input_line = NULL;
+	struct PoEntry* newchunk   = NULL;
+	struct PoEntry* entry	   = NULL;
+	struct PoEntry* current	   = NULL;
+	FILE* input		   = NULL;
+	char* input_line	   = NULL;
 	ParseState pstate	   = PS_NONE;
 	int has_plural_msgid	   = 0;
-	ssize_t input_size	    = limits[L_LINE_MAX];
+	ssize_t input_size	   = limits[L_LINE_MAX];
 	ssize_t non_obsolete_sofar = 0;
 	/* lines to tolerate the lack of msgid */
 	int msgid_counter = PO_DETECTION_LINES;
 	int msgstr_index  = -1;
+	int possible_nulbyte;
 
 	assert(dstate != NULL);
 	input_line = malloc(input_size);
@@ -455,27 +456,44 @@ load_file(struct DrawState* dstate, long* lineno, long* col)
 	dstate->action = ACT_READING;
 	redraw_status(dstate);
 
+	possible_nulbyte = 0;
+	*lineno		 = 1;
 	while (!feof(input))
 	{
-		char* eol = NULL;
-		int result;
+		char* eol     = NULL;
+		char* nulbyte = NULL;
 
 		if (!fgets(input_line, input_size, input))
 			continue;
 
+		if (possible_nulbyte)
+		{
+			/* Definitely not EOF, NUL byte confirmed */
+			dstate->error = ERR_NOTPO_FILE;
+			fclose(input);
+			free(input_line);
+			return 1;
+		}
 		eol = strchr(input_line, '\n');
 		if (eol)
 			*eol = 0;
+		nulbyte = strchr(input_line, 0);
 
-		result = parse_po_line(input_line, dstate, &entry, &pstate, col,
+		/* We don't know if the NUL byte not being equal to
+		 * (possibly NULL) eol is the result of EOF or not
+		 */
+		if (nulbyte && (nulbyte != eol))
+			possible_nulbyte = 1;
+
+		parse_po_line(input_line, dstate, &entry, &pstate, col,
 			&msgstr_index);
 
 		if (entry && entry->msgid_plural)
 			has_plural_msgid = 1;
 
-		switch(result)
+		switch (dstate->error)
 		{
-		case PARSE_ERR_NONE:
+		case ERR_NONE:
 			if (pstate & PS_MSGID)
 				msgid_counter = -1;
 			else if (msgid_counter != -1)
@@ -487,13 +505,14 @@ load_file(struct DrawState* dstate, long* lineno, long* col)
 				return 1;
 			}
 			break;
-		case PARSE_ERR_SYNTAX:
-			dstate->error = ERR_SYNTAX;
-			return 1;
+		case ERR_SYNTAX:
+			/* Allow for ERR_NOTPO_FILE to override ERR_SYNTAX */
+			break;
 		default:
 			goto load_file_alloc_error;
 		}
-		(*lineno)++;
+		if (!dstate->error)
+			(*lineno)++;
 	}
 
 	/* when the .po file is not a standard text file (no LF at end) */
@@ -1069,49 +1088,52 @@ save_file_cleanup:
 	return result;
 }
 
+#define INC_COL()                   \
+	do                          \
+	{                           \
+		if (!dstate->error) \
+			(*col)++;   \
+	} while (0)
+
 int
 parse_po_line(const char* line, struct DrawState* dstate,
 	struct PoEntry** current_entry, ParseState* pstate, long* col,
 	int* msgstr_index)
 {
-	size_t i	  = 0;
-	*col		  = 1;
 	const char* pline = line;
-	char token[MAXBUFLINE];
-	*token	     = 0;
-	char* ptoken = token;
+	char* token	  = NULL;
+	char* ptoken	  = NULL;
+	size_t i	  = 0;
+	size_t len;
 
 	assert(line != NULL);
-	/*if (!line)
-		return PARSE_ERR_INVAL;*/
 
-	if (starts_with(line, "msgid "))
-	{
-		size_t len = strlen("msgid ");
-		*pstate |= PS_MSGID;
-		pline += len;
-		*col += len;
-	}
-	else if (starts_with(line, "msgid_plural "))
+	*col = 1;
+
+	token = malloc(limits[L_LINE_MAX]);
+	if (!token)
+		goto parse_po_line_alloc_error;
+	*token = 0;
+	ptoken = token;
+
+	if (starts_with(line, "msgid_plural"))
 	{
-		size_t len = strlen("msgid_plural ");
+		len = strlen("msgid_plural");
 		*pstate &= ~PS_MSGID;
 		*pstate |= PS_MSGID_PLURAL;
 		pline += len;
 		*col += len;
 	}
-	else if (starts_with(line, "msgstr "))
+	else if (starts_with(line, "msgid"))
 	{
-		size_t len = strlen("msgstr ");
-		*pstate &= ~PS_MSGID;
-		*pstate &= ~PS_MSGID_PLURAL;
-		*pstate |= PS_MSGSTR;
+		len = strlen("msgid");
+		*pstate |= PS_MSGID;
 		pline += len;
 		*col += len;
 	}
 	else if (starts_with(line, "msgstr["))
 	{
-		size_t len = strlen("msgstr[");
+		len = strlen("msgstr[");
 		*pstate &= ~PS_MSGID;
 		*pstate &= ~PS_MSGID_PLURAL;
 		*pstate |= PS_MSGSTR_ARRAY;
@@ -1131,7 +1153,19 @@ parse_po_line(const char* line, struct DrawState* dstate,
 			(*col) += 2;
 		}
 		else
-			return PARSE_ERR_SYNTAX;
+		{
+			dstate->error = ERR_SYNTAX;
+			goto parse_po_cleanup;
+		}
+	}
+	else if (starts_with(line, "msgstr"))
+	{
+		len = strlen("msgstr");
+		*pstate &= ~PS_MSGID;
+		*pstate &= ~PS_MSGID_PLURAL;
+		*pstate |= PS_MSGSTR;
+		pline += len;
+		*col += len;
 	}
 	else if (*line == 0)
 	{
@@ -1141,121 +1175,155 @@ parse_po_line(const char* line, struct DrawState* dstate,
 		*pstate &= ~PS_MSGSTR_ARRAY;
 	}
 	else if (*line != ' ' && *line != '\t' && *line != '"' && *line != '#')
-		return PARSE_ERR_SYNTAX;
+	{
+		dstate->error = ERR_SYNTAX;
+		goto parse_po_cleanup;
+	}
 
 	while (*pline && i < strlen(line))
 	{
 		switch (*pline)
 		{
+		case ' ':
+		case '\t':
+			if (*pstate & PS_COMMENT)
+			{
+				*ptoken++ = *pline++;
+				INC_COL();
+			}
+			else if (*pstate & PS_STRING)
+			{
+				*ptoken++ = *pline++;
+				INC_COL();
+			}
+			else
+			{
+				pline++;
+				INC_COL();
+			}
+			break;
 		case '#':
-			if (*col == 1)
+			if (pline - line + 1 == 1)
 			{
 				*pstate |= PS_COMMENT_TRAN;
 				pline++;
-				(*col)++;
+				INC_COL();
 			}
-			else
+			else if (*pstate & (PS_STRING | PS_COMMENT))
 			{
 				*ptoken++ = *pline++;
-				(*col)++;
+				INC_COL();
+			}
+			else
+			{
+				dstate->error = ERR_SYNTAX;
+				goto parse_po_cleanup;
 			}
 			break;
 		case '.':
-			if (*col == 2 && *pstate & PS_COMMENT_TRAN)
+			if (pline - line + 1 == 2 && *pstate & PS_COMMENT_TRAN)
 			{
 				*pstate &= ~PS_COMMENT_TRAN;
 				*pstate |= PS_COMMENT_EXTR;
 				pline++;
-				(*col)++;
-				if (*pline == ' ')
-				{
-					pline++;
-					(*col)++;
-				}
-				else if (*pline)
-					return PARSE_ERR_SYNTAX;
+				INC_COL();
 			}
 			else
 			{
 				*ptoken++ = *pline++;
-				(*col)++;
+				INC_COL();
 			}
 			break;
 		case ',':
-			if (*col == 2 && *pstate & PS_COMMENT_TRAN)
+			if (pline - line + 1 == 2 && *pstate & PS_COMMENT_TRAN)
 			{
 				*pstate &= ~PS_COMMENT_TRAN;
 				*pstate |= PS_COMMENT_FLAG;
 				pline++;
-				(*col)++;
-				if (*pline == ' ')
-				{
-					pline++;
-					(*col)++;
-				}
-				else if (*pline)
-					return PARSE_ERR_SYNTAX;
+				INC_COL();
 			}
 			else
 			{
 				*ptoken++ = *pline++;
-				(*col)++;
+				INC_COL();
 			}
 			break;
 		case ':':
-			if (*col == 2 && *pstate & PS_COMMENT_TRAN)
+			if (pline - line + 1 == 2 && *pstate & PS_COMMENT_TRAN)
 			{
 				*pstate &= ~PS_COMMENT_TRAN;
 				*pstate |= PS_COMMENT_REFN;
 				pline++;
-				(*col)++;
-				if (*pline == ' ')
-				{
-					pline++;
-					(*col)++;
-				}
-				else if (*pline)
-					return PARSE_ERR_SYNTAX;
+				INC_COL();
 			}
 			else
 			{
 				*ptoken++ = *pline++;
-				(*col)++;
+				INC_COL();
 			}
 			break;
 		case '~':
-			if (*col == 2 && *pstate & PS_COMMENT_TRAN)
+			if (pline - line + 1 == 2 && (*pstate & PS_COMMENT_TRAN))
 			{
 				*pstate &= ~PS_COMMENT_TRAN;
 				*pstate |= PS_COMMENT_OBSO;
 				pline++;
-				(*col)++;
+				INC_COL();
 			}
 			else
 			{
 				*ptoken++ = *pline++;
-				(*col)++;
+				INC_COL();
+			}
+			break;
+		case '\\':
+			if (*pstate & PS_COMMENT)
+			{
+				pline++;
+				INC_COL();
+			}
+			else if ((*pstate & PS_STRING) && *(pstate + 1))
+			{
+				*ptoken++ = *pline++;
+				*ptoken++ = *pline++;
+				INC_COL();
 			}
 			break;
 		case '"':
-			/* skip beginning and end quote */
-			if ((ptoken == token || *(pline + 1) == 0)
-				&& !(*pstate
-					& (PS_COMMENT_TRAN | PS_COMMENT_EXTR
-						| PS_COMMENT_OBSO)))
+			if (*pstate & PS_COMMENT)
+			{
+				pline++;
+				INC_COL();
+			}
+			else if (*pstate & PS_STRING)
 			{
+				*pstate &= ~PS_STRING;
 				pline++;
-				(*col)++;
+				INC_COL();
 			}
 			else
 			{
-				*ptoken++ = *pline++;
-				(*col)++;
+				*pstate |= PS_STRING;
+				pline++;
+				INC_COL();
 			}
 			break;
 		default:
-			*ptoken++ = *pline++;
-			(*col)++;
+			if (*pstate & PS_COMMENT)
+			{
+				*ptoken++ = *pline++;
+				INC_COL();
+			}
+			else if (*pstate & PS_STRING)
+			{
+				*ptoken++ = *pline++;
+				INC_COL();
+			}
+			else
+			{
+				dstate->error = ERR_SYNTAX;
+				goto parse_po_cleanup;
+			}
 		}
 	}
 	*ptoken = 0;
@@ -1273,7 +1341,7 @@ parse_po_line(const char* line, struct DrawState* dstate,
 			newchunk = realloc(dstate->entries,
 				dstate->msgid_size * sizeof(struct PoEntry));
 			if (!newchunk)
-				return PARSE_ERR_ALLOC;
+				goto parse_po_line_alloc_error;
 			dstate->entries = newchunk;
 			*current_entry	= dstate->entries + dstate->msgid_size
 				- ALLOC_DELTA;
@@ -1292,7 +1360,7 @@ parse_po_line(const char* line, struct DrawState* dstate,
 				for (int i = 0; i < dstate->nplurals; i++)
 					if (!set_msgstr(*current_entry, NULL, i,
 						    0, dstate))
-						return PARSE_ERR_ALLOC;
+						goto parse_po_line_alloc_error;
 		}
 		else if (*pstate & PS_MSGSTR)
 		{
@@ -1309,13 +1377,13 @@ parse_po_line(const char* line, struct DrawState* dstate,
 			}
 
 			if (!set_msgstr(*current_entry, token, 0, 1, dstate))
-				return PARSE_ERR_ALLOC;
+				goto parse_po_line_alloc_error;
 		}
 		else if (*pstate & PS_MSGSTR_ARRAY)
 		{
 			if (!set_msgstr(*current_entry, token, *msgstr_index, 1,
 				    dstate))
-				return PARSE_ERR_ALLOC;
+				goto parse_po_line_alloc_error;
 		}
 		else if (*pstate & PS_COMMENT_TRAN)
 		{
@@ -1323,7 +1391,7 @@ parse_po_line(const char* line, struct DrawState* dstate,
 			size_t len	    = strlen(token) + 1;
 			u32_token	    = calloc(len, sizeof(uint32_t));
 			if (!u32_token)
-				return PARSE_ERR_ALLOC;
+				goto parse_po_line_alloc_error;
 			u8_string_to_unicode(u32_token, token, len);
 			u32_add_comment(*current_entry, u32_token);
 			free(u32_token);
@@ -1334,7 +1402,7 @@ parse_po_line(const char* line, struct DrawState* dstate,
 			size_t len	    = strlen(token) + 1;
 			u32_token	    = calloc(len, sizeof(uint32_t));
 			if (!u32_token)
-				return PARSE_ERR_ALLOC;
+				goto parse_po_line_alloc_error;
 			u8_string_to_unicode(u32_token, token, len);
 			u32_add_trans_comment(*current_entry, u32_token);
 			free(u32_token);
@@ -1374,11 +1442,7 @@ parse_po_line(const char* line, struct DrawState* dstate,
 				{
 					/* Let this slip, ignore everything
 					 * instead; see po.h for rationale */
-					/*return PARSE_ERR_SYNTAX;*/
-
-					while (*ptoken && *ptoken != ' '
-						&& *ptoken != ','
-						&& *ptoken != '\t')
+					while (*ptoken)
 						ptoken++;
 				}
 			}
@@ -1391,7 +1455,7 @@ parse_po_line(const char* line, struct DrawState* dstate,
 			size_t len	    = strlen(token) + 1;
 			u32_token	    = calloc(len, sizeof(uint32_t));
 			if (!u32_token)
-				return PARSE_ERR_ALLOC;
+				goto parse_po_line_alloc_error;
 			u8_string_to_unicode(u32_token, token, len);
 			if (u32_add_comment(*current_entry, u32_token))
 				(*current_entry)->obsolete = 1;
@@ -1405,10 +1469,14 @@ parse_po_line(const char* line, struct DrawState* dstate,
 		u32_add_trans_comment(*current_entry, (uint32_t*)L"");
 
 	/* one-line states */
-	*pstate &= ~(PS_COMMENT_TRAN | PS_COMMENT_EXTR | PS_COMMENT_FLAG
-		| PS_COMMENT_REFN | PS_COMMENT_OBSO);
+	*pstate &= ~PS_COMMENT;
 
-	return PARSE_ERR_NONE;
+parse_po_cleanup:
+	return !!dstate->error;
+
+parse_po_line_alloc_error:
+	dstate->error = ERR_CANT_ALLOC;
+	return 1;
 }
 
 void
diff --git a/po.h b/po.h
@@ -11,11 +11,11 @@
 /* If msgid/msgstr is longer, it will be saved as multiple lines */
 #define SAVE_MULTILINE_TRESHOLD 70
 
-#define LOAD_PARSE_ERR(e) (LOAD_ERR_PARSE_PREFIX + (e))
+// #define LOAD_PARSE_ERR(e) (LOAD_ERR_PARSE_PREFIX + (e))
 
 struct DrawState;
 
-enum {
+/*enum {
 	PARSE_ERR_NONE	 = 0,
 	PARSE_ERR_INVAL	 = 1,
 	PARSE_ERR_ALLOC	 = 2,
@@ -31,7 +31,7 @@ enum {
 	LOAD_ERR_P_INVAL	= LOAD_PARSE_ERR(PARSE_ERR_INVAL),
 	LOAD_ERR_P_ALLOC	= LOAD_PARSE_ERR(PARSE_ERR_ALLOC),
 	LOAD_ERR_P_SYNTAX	= LOAD_PARSE_ERR(PARSE_ERR_SYNTAX)
-};
+};*/
 
 enum {
 	SAVE_ERR_NONE		= 0,
@@ -54,9 +54,13 @@ typedef enum {
 	PS_MSGID	= 1 << 5,
 	PS_MSGID_PLURAL = 1 << 6,
 	PS_MSGSTR	= 1 << 7,
-	PS_MSGSTR_ARRAY = 1 << 8
+	PS_MSGSTR_ARRAY = 1 << 8,
+	PS_STRING	= 1 << 9 /* string literal */
 } ParseState;
 
+#define PS_COMMENT                                            \
+	((PS_COMMENT_TRAN | PS_COMMENT_EXTR | PS_COMMENT_FLAG \
+		| PS_COMMENT_REFN | PS_COMMENT_OBSO))
 /* Complete list at:
  *
  * https://www.gnu.org/software/gettext/manual/html_node/PO-Files.html
@@ -97,40 +101,35 @@ struct PoEntry {
 	size_t ref_comment_lines;
 };
 
-char* format_flags(char* buffer, const size_t max,
-	const struct PoEntry* entry, const int draw_box);
+char* format_flags(char* buffer, const size_t max, const struct PoEntry* entry,
+	const int draw_box);
 void init_po_entry(struct PoEntry* entry);
 void free_po_entry(struct PoEntry* entry);
 struct PoEntry* set_flags(struct PoEntry* entry, const Flags flags);
-struct PoEntry* set_msgid(struct PoEntry* entry, const char* msgid,
-	int append);
+struct PoEntry* set_msgid(struct PoEntry* entry, const char* msgid, int append);
 struct PoEntry* set_msgid_plural(struct PoEntry* entry,
 	const char* msgid_plural, int append);
 struct PoEntry* set_plural_forms(struct PoEntry* entry,
 	const char* plural_forms, int append);
 struct PoEntry* set_msgstr(struct PoEntry* entry, const char* msgstr,
 	int msgstr_index, int append, struct DrawState* state);
-struct PoEntry* u32_set_msgid(struct PoEntry* entry,
-	const uint32_t* msgid, size_t msgid_size, int append);
+struct PoEntry* u32_set_msgid(struct PoEntry* entry, const uint32_t* msgid,
+	size_t msgid_size, int append);
 struct PoEntry* u32_set_msgid_plural(struct PoEntry* entry,
 	const uint32_t* msgid_plural, size_t msgid_plural_size, int append);
 struct PoEntry* u32_set_plural_forms(struct PoEntry* entry,
 	const uint32_t* plural_forms, int append);
-struct PoEntry* u32_set_msgstr(struct PoEntry* entry,
-	const uint32_t* msgstr, int msgstr_index, int append,
-	struct DrawState* state);
-struct PoEntry* u32_add_comment(struct PoEntry* entry,
-	const uint32_t* comment);
+struct PoEntry* u32_set_msgstr(struct PoEntry* entry, const uint32_t* msgstr,
+	int msgstr_index, int append, struct DrawState* state);
+struct PoEntry* u32_add_comment(struct PoEntry* entry, const uint32_t* comment);
 struct PoEntry* u32_add_trans_comment(struct PoEntry* entry,
 	const uint32_t* trans_comment);
-struct PoEntry* add_ref_comment(struct PoEntry* entry,
-	const char* comment);
+struct PoEntry* add_ref_comment(struct PoEntry* entry, const char* comment);
 int load_file(struct DrawState* state, long* lineno, long* col);
 int save_file(struct DrawState* state);
 int parse_po_line(const char* line, struct DrawState* dstate,
 	struct PoEntry** current_entry, ParseState* pstate, long* col,
 	int* msgstr_index);
-void update_statistics(const struct PoEntry* entries,
-	struct DrawState* state, struct PoEntry* entry);
-void update_warning(const uint32_t* msgid, const uint32_t* msgstr,
-	int* warning);
+void update_statistics(const struct PoEntry* entries, struct DrawState* state,
+	struct PoEntry* entry);
+void update_warning(const uint32_t* msgid, const uint32_t* msgstr, int* warning);
diff --git a/poe.c b/poe.c
@@ -2670,10 +2670,19 @@ main_done:
 	if (saved_errno)
 		perror(PROGRAM_NAME);
 
-	if (state.error == ERR_NOTREG_FILE)
+	switch (state.error)
+	{
+	case ERR_NOTREG_FILE:
 		print_error(1, errors[state.error], state.filename);
-	else if (state.error)
+		break;
+	case ERR_SYNTAX:
+		print_error(1, errors[state.error], state.filename, lineno, col);
+		break;
+	case ERR_NONE:
+		break;
+	default:
 		print_error(1, errors[state.error]);
+	}
 
 	return saved_errno
 		? saved_errno