X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=sherlock%2Fxml%2Fxml.h;h=e94588836bf733470c2d008725d08b06fa140252;hb=d5fdccbecd2acde9a6e067b54fcd69b02f31a820;hp=8e416dbc8a775966ac2fe83c66a0227c7408aeb7;hpb=ccf64507b45774b007ab6200036827f1597022d8;p=libucw.git

diff --git a/sherlock/xml/xml.h b/sherlock/xml/xml.h
index 8e416dbc..e9458883 100644
--- a/sherlock/xml/xml.h
+++ b/sherlock/xml/xml.h
@@ -10,13 +10,12 @@
 #ifndef _SHERLOCK_XML_XML_H
 #define _SHERLOCK_XML_XML_H
 
-#include "lib/clists.h"
-#include "lib/slists.h"
-#include "lib/mempool.h"
-#include "lib/fastbuf.h"
+#include "ucw/clists.h"
+#include "ucw/slists.h"
+#include "ucw/mempool.h"
+#include "ucw/fastbuf.h"
 
 struct xml_context;
-struct xml_source;
 struct xml_dtd_entity;
 
 enum xml_error {
@@ -81,18 +80,19 @@ enum xml_flags {
   /* Other parameters */
   XML_VALIDATING =			0x00000100,	/* Validate everything (not fully implemented!) */
   XML_PARSE_DTD =			0x00000200,	/* Enable parsing of DTD */
+  XML_NO_CHARS =			0x00000400,	/* The current element must not contain character data (filled automaticaly if using DTD) */
+  XML_ALLOC_DEFAULT_ATTRS =		0x00000800,	/* Allocate default attribute values so they can be found by XML_ATTR_FOR_EACH */
 
   /* Internals, do not change! */
   XML_EMPTY_ELEM_TAG =			0x00010000,	/* The current element match EmptyElemTag */
   XML_VERSION_1_1 =			0x00020000,	/* XML version is 1.1, otherwise 1.0 */
   XML_HAS_EXTERNAL_SUBSET =		0x00040000,	/* The document contains a reference to external DTD subset */
   XML_HAS_INTERNAL_SUBSET =		0x00080000,	/* The document contains an internal subset */
+  XML_HAS_DTD =	XML_HAS_EXTERNAL_SUBSET | XML_HAS_INTERNAL_SUBSET,
   XML_SRC_EOF =				0x00100000,	/* EOF reached */
   XML_SRC_EXPECTED_DECL =		0x00200000,	/* Just before optional or required XMLDecl/TextDecl */
-  XML_SRC_NEW_LINE =			0x00400000,	/* The last read character is 0xD */
-  XML_SRC_SURROUND =			0x00800000,	/* Surround the text with 0x20 (references to parameter entities) */
-  XML_SRC_DOCUMENT =			0x01000000,	/* The document entity */
-  XML_SRC_EXTERNAL =			0x02000000,	/* An external entity */
+  XML_SRC_DOCUMENT =			0x00400000,	/* The document entity */
+  XML_SRC_EXTERNAL =			0x00800000,	/* An external entity */
 };
 
 enum xml_node_type {
@@ -133,6 +133,28 @@ struct xml_attr {
   void *user;						/* User-defined (initialized to NULL) */
 };
 
+#define XML_BUF_SIZE 32                                 /* At least 8 -- hardcoded */
+
+struct xml_source {
+  struct xml_source *next;				/* Link list of pending fastbufs (xml_context.sources) */
+  struct fastbuf *fb;					/* Source fastbuf */
+  struct fastbuf *wrapped_fb;				/* Original wrapped fastbuf (needed for cleanup) */
+  struct fastbuf wrap_fb;				/* Fbmem wrapper */
+  u32 buf[2 * XML_BUF_SIZE];				/* Read buffer with Unicode values and categories */
+  u32 *bptr, *bstop;					/* Current state of the buffer */
+  uns row;						/* File position */
+  char *expected_encoding;				/* Initial encoding before any transformation has been made (expected in XMLDecl/TextDecl) */
+  char *fb_encoding;					/* Encoding of the source fastbuf */
+  char *decl_encoding;					/* Encoding read from the XMLDecl/TextDecl */
+  uns refill_cat1;					/* Character categories, which should be directly passed to the buffer */
+  uns refill_cat2;					/* Character categories, which should be processed as newlines (possibly in some built-in
+							   sequences) */
+  void (*refill)(struct xml_context *ctx);		/* Callback to decode source characters to the buffer */
+  unsigned short *refill_in_to_x;			/* Libcharset input table */
+  uns saved_depth;					/* Saved ctx->depth */
+  uns pending_0xd;					/* The last read character is 0xD */
+};
+
 struct xml_context {
   /* Error handling */
   char *err_msg;					/* Last error message */
@@ -174,6 +196,7 @@ struct xml_context {
   void (*h_chars)(struct xml_context *ctx);		/* Called after some characters (only with XML_REPORT_CHARS) */
   void (*h_block)(struct xml_context *ctx, char *text, uns len);	/* Called for each continuous block of characters not reported by h_cdata() (only with XML_REPORT_CHARS) */
   void (*h_cdata)(struct xml_context *ctx, char *text, uns len);	/* Called for each CDATA section (only with XML_REPORT_CHARS) */
+  void (*h_ignorable)(struct xml_context *ctx, char *text, uns len);	/* Called for ignorable whitespace (content in tags without #PCDATA) */
   void (*h_dtd_start)(struct xml_context *ctx);		/* Called just after the DTD structure is initialized */
   void (*h_dtd_end)(struct xml_context *ctx);		/* Called after DTD subsets subsets */
   struct xml_dtd_entity *(*h_find_entity)(struct xml_context *ctx, char *name);		/* Called when needed to resolve a general entity */
@@ -191,11 +214,6 @@ struct xml_context {
   struct xml_dtd *dtd;					/* The DTD structure (or NULL) */
   uns state;						/* Current state for the PULL interface (XML_STATE_x) */
   uns pull;						/* Parameters for the PULL interface (XML_PULL_x) */
-
-  void (*start_entity)(struct xml_context *ctx);
-  void (*end_entity)(struct xml_context *ctx);
-  void (*notation_decl)(struct xml_context *ctx);
-  void (*unparsed_entity_decl)(struct xml_context *ctx);
 };
 
 /* Initialize XML context */
@@ -204,7 +222,7 @@ void xml_init(struct xml_context *ctx);
 /* Clean up all internal structures */
 void xml_cleanup(struct xml_context *ctx);
 
-/* Reuse XML context */
+/* Reuse XML context, equivalent to xml_cleanup() and xml_init() */
 void xml_reset(struct xml_context *ctx);
 
 /* Add XML source (fastbuf will be automatically closed) */
@@ -216,12 +234,21 @@ uns xml_parse(struct xml_context *ctx);
 /* Parse with the PUSH interface, return XML_STATE_x (zero on EOF or fatal error) */
 uns xml_next(struct xml_context *ctx);
 
+/* Equivalent to xml_next, but with temporarily changed ctx->pull value */
+uns xml_next_state(struct xml_context *ctx, uns pull);
+
+/* May be called on XML_STATE_STAG to skip it's content; can return XML_STATE_ETAG or XML_STATE_EOF on fatal error */
+uns xml_skip_element(struct xml_context *ctx);
+
 /* Returns the current row number in the document entity */
 uns xml_row(struct xml_context *ctx);
 
 /* Finds a given attribute value in a XML_NODE_ELEM node */
 struct xml_attr *xml_attr_find(struct xml_context *ctx, struct xml_node *node, char *name);
 
+/* Similar to xml_attr_find, but it deals also with default values */
+char *xml_attr_value(struct xml_context *ctx, struct xml_node *node, char *name);
+
 /* The default value of h_find_entity(), knows <, >, &, ' and " */
 struct xml_dtd_entity *xml_def_find_entity(struct xml_context *ctx, char *name);
 
@@ -231,4 +258,15 @@ void xml_def_resolve_entity(struct xml_context *ctx, struct xml_dtd_entity *ent)
 /* Remove leading/trailing spaces and replaces sequences of spaces to a single space character (non-CDATA attribute normalization) */
 uns xml_normalize_white(struct xml_context *ctx, char *value);
 
+/* Merge character contents of a given element to a single string (not recursive) */
+char *xml_merge_chars(struct xml_context *ctx, struct xml_node *node, struct mempool *pool);
+
+/* Merge character contents of a given subtree to a single string */
+char *xml_merge_dom_chars(struct xml_context *ctx, struct xml_node *node, struct mempool *pool);
+
+/* Public part of error handling */
+void xml_warn(struct xml_context *ctx, const char *format, ...);
+void xml_error(struct xml_context *ctx, const char *format, ...);
+void NONRET xml_fatal(struct xml_context *ctx, const char *format, ...);
+
 #endif