XML: Bugfix in processing of multi-byte newlines.

[libucw.git] / sherlock / xml / xml.h

diff --git a/sherlock/xml/xml.h b/sherlock/xml/xml.h
index 8e416dbc8a775966ac2fe83c66a0227c7408aeb7..ac9ebefb70f8a88ab6f1056326dd876eb0e44002 100644 (file)
--- a/sherlock/xml/xml.h
+++ b/sherlock/xml/xml.h
@@ -16,7 +16,6 @@
 #include "lib/fastbuf.h"
 
 struct xml_context;
-struct xml_source;
 struct xml_dtd_entity;
 
 enum xml_error {
@@ -89,10 +88,8 @@ enum xml_flags {
   XML_HAS_INTERNAL_SUBSET =            0x00080000,     /* The document contains an 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 +130,28 @@ 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 */
@@ -191,11 +210,6 @@ 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 +218,7 @@ 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) */
@@ -231,4 +245,9 @@ 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);
 
+/* 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