Restore binary compat, more HTML stuff, allow stdin input, Daniel.

ChangeLog

@@ -1,3 +1,9 @@
+Wed Jul  7 00:25:42 CEST 1999 Daniel Veillard <Daniel.Veillard@w3.org>
+
+	* parser.h : Oops removed the binary compatibility problem
+	* HTMLparser.[ch], HTMLtree.h : More work on the HTML parse/dump
+	* parser.c, HTMLparser.c: applied patches for reading from stdin
+
 Mon Jul  5 18:45:31 CEST 1999 Daniel Veillard <Daniel.Veillard@w3.org>
 
 	* parser.c, entities.c, valid.c: cleanup bug #1591

HTMLparser.h

@@ -10,6 +10,9 @@
 #define __HTML_PARSER_H__
 #include "parser.h"
 
+/*
+ * Most of the back-end structures from XML and HTML are shared
+ */
 typedef xmlParserCtxt htmlParserCtxt;
 typedef xmlParserCtxtPtr htmlParserCtxtPtr;
 typedef xmlParserNodeInfo htmlParserNodeInfo;
@@ -20,7 +23,33 @@ typedef xmlParserInputPtr htmlParserInputPtr;
 typedef xmlDocPtr htmlDocPtr;
 typedef xmlNodePtr htmlNodePtr;
 
-xmlEntityPtr htmlParseEntityRef(htmlParserCtxtPtr ctxt);
+/*
+ * Internal description of an HTML element
+ */
+typedef struct htmlElemDesc {
+    const CHAR *name;	/* The tag name */
+    int startTag;       /* Whether the start tag can be implied */
+    int endTag;         /* Whether the end tag can be implied */
+    int empty;          /* Is this an empty element ? */
+    int depr;           /* Is this a deprecated element ? */
+    int dtd;            /* 1: only in Loose DTD, 2: only Frameset one */
+    const char *desc;   /* the description */
+} htmlElemDesc, *htmlElemDescPtr;
+
+/*
+ * Internal description of an HTML entity
+ */
+typedef struct htmlEntityDesc {
+    int value;		/* the UNICODE value for the character */
+    const CHAR *name;	/* The entity name */
+    const char *desc;   /* the description */
+} htmlEntityDesc, *htmlEntityDescPtr;
+
+/*
+ * There is only few public functions.
+ */
+htmlEntityDescPtr
+htmlParseEntityRef(htmlParserCtxtPtr ctxt, CHAR **str);
 int htmlParseCharRef(htmlParserCtxtPtr ctxt);
 void htmlParseElement(htmlParserCtxtPtr ctxt);
 

Makefile.am

@@ -23,6 +23,7 @@ libxml_la_SOURCES = \
 		HTMLparser.c \
 		debugXML.c \
 		tree.c \
+		HTMLtree.c \
 		valid.c
 
 xmlincdir = $(includedir)/gnome-xml
@@ -30,6 +31,7 @@ xmlinc_HEADERS = \
 		entities.h \
 		encoding.h \
 		parser.h \
+		HTMLparser.h \
 		parserInternals.h \
 		debugXML.h \
 		xml-error.h \

SAX.c

@@ -180,8 +180,14 @@ resolveEntity(void *ctx, const CHAR *publicId, const CHAR *systemId)
      * TODO : not 100% sure that the appropriate handling in that case.
      */
     if (systemId != NULL) {
+        if (!xmlStrncmp(systemId, "http://", 7)) {
+	    /* !!!!!!!!! TODO */
+	} else if (!xmlStrncmp(systemId, "ftp://", 6)) {
+	    /* !!!!!!!!! TODO */
+	} else {
 	    return(xmlNewInputFromFile(ctxt, systemId));
 	}
+    }
     return(NULL);
 }
 

include/libxml/HTMLparser.h

@@ -10,6 +10,9 @@
 #define __HTML_PARSER_H__
 #include "parser.h"
 
+/*
+ * Most of the back-end structures from XML and HTML are shared
+ */
 typedef xmlParserCtxt htmlParserCtxt;
 typedef xmlParserCtxtPtr htmlParserCtxtPtr;
 typedef xmlParserNodeInfo htmlParserNodeInfo;
@@ -20,7 +23,33 @@ typedef xmlParserInputPtr htmlParserInputPtr;
 typedef xmlDocPtr htmlDocPtr;
 typedef xmlNodePtr htmlNodePtr;
 
-xmlEntityPtr htmlParseEntityRef(htmlParserCtxtPtr ctxt);
+/*
+ * Internal description of an HTML element
+ */
+typedef struct htmlElemDesc {
+    const CHAR *name;	/* The tag name */
+    int startTag;       /* Whether the start tag can be implied */
+    int endTag;         /* Whether the end tag can be implied */
+    int empty;          /* Is this an empty element ? */
+    int depr;           /* Is this a deprecated element ? */
+    int dtd;            /* 1: only in Loose DTD, 2: only Frameset one */
+    const char *desc;   /* the description */
+} htmlElemDesc, *htmlElemDescPtr;
+
+/*
+ * Internal description of an HTML entity
+ */
+typedef struct htmlEntityDesc {
+    int value;		/* the UNICODE value for the character */
+    const CHAR *name;	/* The entity name */
+    const char *desc;   /* the description */
+} htmlEntityDesc, *htmlEntityDescPtr;
+
+/*
+ * There is only few public functions.
+ */
+htmlEntityDescPtr
+htmlParseEntityRef(htmlParserCtxtPtr ctxt, CHAR **str);
 int htmlParseCharRef(htmlParserCtxtPtr ctxt);
 void htmlParseElement(htmlParserCtxtPtr ctxt);
 

include/libxml/tree.h

@@ -230,8 +230,6 @@ typedef struct xmlDoc {
     char           *name;	/* name/filename/URI of the document */
     const CHAR     *version;	/* the XML version string */
     const CHAR     *encoding;   /* encoding, if any */
-    const CHAR     *ID;		/* the HTML version */
-    const CHAR     *DTD;   	/* the HTML dtd URI */
     int             compression;/* level of zlib compression */
     int             standalone; /* standalone document (no external refs) */
     struct xmlDtd  *intSubset;	/* the document internal subset */

parser.c

@@ -264,28 +264,36 @@ xmlNewInputFromFile(xmlParserCtxtPtr ctxt, const char *filename) {
 #endif
     int res;
     int len;
+    int cnt;
     struct stat buf;
-    char *buffer;
+    char *buffer, *nbuf;
     xmlParserInputPtr inputStream;
     /* xmlCharEncoding enc; */
 
-    res = stat(filename, &buf);
+#define MINLEN 40000
-    if (res < 0) return(NULL);
 
+    if (strcmp(filename,"-") == 0) {
 #ifdef HAVE_ZLIB_H
-    len = (buf.st_size * 8) + 1000;
+        input = gzdopen (fileno(stdin), "r");
-retry_bigger:
+        if (input == NULL) {
-    buffer = malloc(len);
+            fprintf (stderr, "Cannot read from stdin\n");
-#else
+            perror ("gzdopen failed");
-    len = buf.st_size + 100;
-    buffer = malloc(len);
-#endif
-    if (buffer == NULL) {
-	perror("malloc");
             return(NULL);
 	}
-
+#else
-    memset(buffer, 0, len);
+#ifdef WIN32
+        input = -1;
+#else
+        input = fileno(stdin);
+#endif
+        if (input < 0) {
+            fprintf (stderr, "Cannot read from stdin\n");
+            perror ("open failed");
+        return(NULL);
+    }
+#endif
+		len = MINLEN;
+    } else {
 #ifdef HAVE_ZLIB_H
 	input = gzopen (filename, "r");
 	if (input == NULL) {
@@ -305,10 +313,41 @@ retry_bigger:
 	    return(NULL);
 	}
 #endif
+	res = stat(filename, &buf);
+	if (res < 0) 
+		return(NULL);
+	len = buf.st_size+1;
+	if (len < MINLEN)
+		len = MINLEN;
+    }
+    buffer = (char *)malloc(len*sizeof(char));
+    if (buffer == NULL) {
+		fprintf (stderr, "Cannot malloc\n");
+		perror ("malloc failed");
+		return(NULL);
+    }
+
+    cnt = 0;
 #ifdef HAVE_ZLIB_H
-    res = gzread(input, buffer, len);
+    while(!gzeof(input)) {
 #else
-    res = read(input, buffer, buf.st_size);
+    while(1) {
+#endif
+	if (cnt >= len) {
+	    len *= 2;
+	    nbuf =  (char *)realloc(buffer,len*sizeof(char));
+	    if (nbuf == NULL) {
+		    fprintf(stderr,"Cannot realloc\n");
+		    free(buffer);
+		    perror ("realloc failed");
+		    return(NULL);
+	    }
+	    buffer = nbuf;
+	}
+#ifdef HAVE_ZLIB_H
+    	res = gzread(input, &buffer[cnt], len-cnt);
+#else
+	res = read(input, &buffer[cnt], len-cnt);
 #endif
 	if (res < 0) {
 	    fprintf (stderr, "Cannot read file %s :\n", filename);
@@ -319,19 +358,17 @@ retry_bigger:
 #endif
 	    return(NULL);
 	}
+	if (res == 0) 
+	    break;
+	cnt += res;
+    }
 #ifdef HAVE_ZLIB_H
     gzclose(input);
-    if (res >= len) {
-        free(buffer);
-	len *= 2;
-	goto retry_bigger;
-    }
-    buf.st_size = res;
 #else
     close(input);
 #endif
 
-    buffer[buf.st_size] = '\0';
+    buffer[cnt] = '\0';
 
     inputStream = (xmlParserInputPtr) malloc(sizeof(xmlParserInput));
     if (inputStream == NULL) {
@@ -4594,28 +4631,36 @@ xmlCreateFileParserCtxt(const char *filename)
 #endif
     int res;
     int len;
+	int cnt;
     struct stat buf;
-    char *buffer;
+    char *buffer, *nbuf;
     xmlParserInputPtr inputStream;
     xmlCharEncoding enc;
 
-    res = stat(filename, &buf);
+#define MINLEN 40000
-    if (res < 0) return(NULL);
 
+	if (strcmp(filename,"-") == 0) {
 #ifdef HAVE_ZLIB_H
-    len = (buf.st_size * 8) + 1000;
+        input = gzdopen (fileno(stdin), "r");
-retry_bigger:
+        if (input == NULL) {
-    buffer = malloc(len);
+            fprintf (stderr, "Cannot read from stdin\n");
-#else
+            perror ("gzdopen failed");
-    len = buf.st_size + 100;
-    buffer = malloc(len);
-#endif
-    if (buffer == NULL) {
-	perror("malloc");
             return(NULL);
 	    }
-
+#else
-    memset(buffer, 0, len);
+#ifdef WIN32
+        input = -1;
+#else
+        input = fileno(stdin);
+#endif
+        if (input < 0) {
+            fprintf (stderr, "Cannot read from stdin\n");
+            perror ("open failed");
+        return(NULL);
+    }
+#endif
+		len = MINLEN;
+    } else {
 #ifdef HAVE_ZLIB_H
     input = gzopen (filename, "r");
     if (input == NULL) {
@@ -4635,10 +4680,41 @@ retry_bigger:
 	return(NULL);
     }
 #endif
+		res = stat(filename, &buf);
+		if (res < 0) 
+			return(NULL);
+		len = buf.st_size+1;
+		if (len < MINLEN)
+			len = MINLEN;
+    }
+	buffer = (char *)malloc(len*sizeof(char));
+    if (buffer == NULL) {
+		fprintf (stderr, "Cannot malloc\n");
+		perror ("malloc failed");
+		return(NULL);
+	}
+
+	cnt = 0;
 #ifdef HAVE_ZLIB_H
-    res = gzread(input, buffer, len);
+	while(!gzeof(input)) {
 #else
-    res = read(input, buffer, buf.st_size);
+	while(1) {
+#endif
+		if (cnt == len) {
+			len *= 2;
+			nbuf =  (char *)realloc(buffer,len*sizeof(char));
+			if (nbuf == NULL) {
+				fprintf(stderr,"Cannot realloc\n");
+				free(buffer);
+				perror ("realloc failed");
+				return(NULL);
+			}
+			buffer = nbuf;
+		}
+#ifdef HAVE_ZLIB_H
+    	res = gzread(input, &buffer[cnt], len-cnt);
+#else
+   		res = read(input, &buffer[cnt], len-cnt);
 #endif
     if (res < 0) {
         fprintf (stderr, "Cannot read file %s :\n", filename);
@@ -4648,20 +4724,18 @@ retry_bigger:
 	perror ("read failed");
 #endif
 	return(NULL);
+    }
+   		if (res == 0) 
+			break;
+		cnt += res;
 	}
 #ifdef HAVE_ZLIB_H
     gzclose(input);
-    if (res >= len) {
-        free(buffer);
-	len *= 2;
-	goto retry_bigger;
-    }
-    buf.st_size = res;
 #else
     close(input);
 #endif
 
-    buffer[res] = '\0';
+    buffer[cnt] = '\0';
 
     ctxt = (xmlParserCtxtPtr) malloc(sizeof(xmlParserCtxt));
     if (ctxt == NULL) {

tree.c

@@ -387,8 +387,6 @@ xmlNewDoc(const CHAR *version) {
 
     cur->type = XML_DOCUMENT_NODE;
     cur->version = xmlStrdup(version); 
-    cur->ID = NULL;
-    cur->DTD = NULL;
     cur->name = NULL;
     cur->root = NULL; 
     cur->intSubset = NULL;

tree.h

@@ -230,8 +230,6 @@ typedef struct xmlDoc {
     char           *name;	/* name/filename/URI of the document */
     const CHAR     *version;	/* the XML version string */
     const CHAR     *encoding;   /* encoding, if any */
-    const CHAR     *ID;		/* the HTML version */
-    const CHAR     *DTD;   	/* the HTML dtd URI */
     int             compression;/* level of zlib compression */
     int             standalone; /* standalone document (no external refs) */
     struct xmlDtd  *intSubset;	/* the document internal subset */