Working on the decoding code.

git-svn-id: http://svn.pyyaml.org/libyaml/trunk@180 18f92427-320e-0410-9341-c67f048884a3

3 files changed, 365 insertions, 52 deletions

diff --git a/include/yaml/yaml.h b/include/yaml/yaml.h
index c4bd0ba..9ec2c7a 100644
--- a/include/yaml/yaml.h
+++ b/include/yaml/yaml.h
@@ -252,7 +252,7 @@ typedef struct {
  * source.  The handler should write not more than @a size bytes to the @a
  * buffer.  The number of written bytes should be set to the @a length variable.
  *
- * @param[in]   ext         A pointer to an application data specified by
+ * @param[in]   data        A pointer to an application data specified by
  *                          @c yaml_parser_set_read_handler.
  * @param[out]  buffer      The buffer to write the data from the source.
  * @param[in]   size        The size of the buffer.
@@ -262,10 +262,22 @@ typedef struct {
  * the returned value should be @c 0.  On EOF, the handler should set the
  * @a length to @c 0 and return @c 1.
  */
-typedef int yaml_read_handler_t(void *ext, unsigned char *buffer, size_t size,
+
+typedef int yaml_read_handler_t(void *data, unsigned char *buffer, size_t size,
         size_t *size_read);
 
 /**
+ * This structure holds a string input specified by
+ * @c yaml_parser_set_input_string.
+ */
+
+typedef struct {
+    unsigned char *start;
+    unsigned char *end;
+    unsigned char *current;
+} yaml_string_input_t;
+
+/**
  * The parser structure.
  *
  * All members are internal.  Manage the structure using the @c yaml_parser_
@@ -279,7 +291,7 @@ typedef struct {
      * @{
      */
 
-    error_type_t error;
+    yaml_error_type_t error;
 
     /**
      * @}
@@ -302,23 +314,23 @@ typedef struct {
     /** The pointer to the beginning of the working buffer. */
     yaml_char_t *buffer;
 
-    /** The size of the buffer (in bytes). */
-    size_t buffer_size;
+    /** The pointer to the end of the working buffer. */
+    yaml_char_t *buffer_end;
 
     /** The pointer to the current character in the working buffer. */
-    yaml_char_t *buffer_pointer;
+    yaml_char_t *pointer;
 
-    /** The number of unread characters in the buffer (in characters). */
-    size_t buffer_length;
+    /** The number of unread characters in the working buffer. */
+    size_t unread;
 
-    /** The remaining undecoded characters. */
+    /** The pointer to the beginning of the raw buffer. */
     unsigned char *raw_buffer;
 
-    /** The size of the raw buffer (in bytes). */
-    size_t raw_buffer_size;
+    /** The pointer to the current character in the raw buffer. */
+    unsigned char *raw_pointer;
 
-    /** Is the application responsible for freeing the raw buffer? */
-    int raw_buffer_foreign;
+    /** The number of unread bytes in the raw buffer. */
+    size_t raw_unread;
 
     /** The input encoding. */
     yaml_encoding_t encoding;
@@ -335,6 +347,9 @@ typedef struct {
     /** The column of the current position (starting from @c 0). */
     size_t column;
 
+    /* String input structure. */
+    yaml_string_input_t string_input;
+
     /**
      * @}
      */
@@ -461,6 +476,18 @@ yaml_realloc(void *ptr, size_t size);
 void
 yaml_free(void *ptr);
 
+/** The size of the raw buffer. */
+
+#define YAML_RAW_BUFFER_SIZE 16384
+
+/**
+ * The size of the buffer.
+ *
+ * We allocate enough space for decoding the whole raw buffer.
+ */
+
+#define YAML_BUFFER_SIZE    (YAML_RAW_BUFFER_SIZE*3)
+
 /** @} */
 
 
diff --git a/src/api.c b/src/api.c
index 0594727..aa183af 100644
--- a/src/api.c
+++ b/src/api.c
@@ -46,11 +46,35 @@ yaml_parser_new(void)
 {
     yaml_parser_t *parser;
 
+    /* Allocate the parser structure. */
+
     parser = yaml_malloc(sizeof(yaml_parser_t));
     if (!parser) return NULL;
 
     memset(parser, 0, sizeof(yaml_parser_t));
 
+    /* Allocate the raw buffer. */
+
+    parser->raw_buffer = yaml_malloc(YAML_RAW_BUFFER_SIZE);
+    if (!parser->raw_buffer) {
+        yaml_free(parser);
+        return NULL;
+    }
+    parser->raw_pointer = parser->raw_buffer;
+    parser->raw_unread = 0;
+
+    /* Allocate the character buffer. */
+
+    parser->buffer = yaml_malloc(YAML_BUFFER_SIZE);
+    if (!parser->buffer) {
+        yaml_free(parser->raw_buffer);
+        yaml_free(parser);
+        return NULL;
+    }
+    parser->buffer_end = parser->buffer;
+    parser->pointer = parser->buffer;
+    parser->unread = 0;
+
     return parser;
 }
 
@@ -64,8 +88,7 @@ yaml_parser_delete(yaml_parser_t *parser)
     assert(parser); /* Non-NULL parser object expected. */
 
     yaml_free(parser->buffer);
-    if (!parser->raw_buffer_foreign)
-        yaml_free(parser->raw_buffer);
+    yaml_free(parser->raw_buffer);
 
     memset(parser, 0, sizeof(yaml_parser_t));
 
@@ -73,14 +96,27 @@ yaml_parser_delete(yaml_parser_t *parser)
 }
 
 /*
- * String read handler (always returns error).
+ * String read handler.
  */
 
 static int
 yaml_string_read_handler(void *data, unsigned char *buffer, size_t size,
         size_t *size_read)
 {
-    *size_read = 0;
+    yaml_string_input_t *input = data;
+
+    if (input->current == input->end) {
+        *size_read = 0;
+        return 1;
+    }
+
+    if (size > (input->end - input->current)) {
+        size = input->end - input->current;
+    }
+
+    memcpy(buffer, input->current, size);
+    input->current += size;
+    *size_read = size;
     return 1;
 }
 
@@ -92,8 +128,8 @@ static int
 yaml_file_read_handler(void *data, unsigned char *buffer, size_t size,
         size_t *size_read)
 {
-    *size_read = fread(buffer, 1, size, (FILE *)ext);
-    return !ferror((FILE *)ext);
+    *size_read = fread(buffer, 1, size, (FILE *)data);
+    return !ferror((FILE *)data);
 }
 
 /*
@@ -105,16 +141,15 @@ yaml_parser_set_input_string(yaml_parser_t *parser,
         unsigned char *input, size_t size)
 {
     assert(parser); /* Non-NULL parser object expected. */
-    assert(!parser->reader); /* You can set the source only once. */
+    assert(!parser->read_handler);  /* You can set the source only once. */
     assert(input);  /* Non-NULL input string expected. */
 
-    parser->read_handler = yaml_string_read_handler;
-    parser->read_handler_data = NULL;
+    parser->string_input.start = input;
+    parser->string_input.current = input;
+    parser->string_input.end = input+size;
 
-    /* We use the input string as a raw (undecoded) buffer. */
-    parser->raw_buffer = input; 
-    parser->raw_buffer_size = size;
-    parser->raw_buffer_foreign = 1;
+    parser->read_handler = yaml_string_read_handler;
+    parser->read_handler_data = &parser->string_input;
 }
 
 /*
@@ -125,7 +160,7 @@ void
 yaml_parser_set_input_file(yaml_parser_t *parser, FILE *file)
 {
     assert(parser); /* Non-NULL parser object expected. */
-    assert(!parser->reader); /* You can set the source only once. */
+    assert(!parser->read_handler);  /* You can set the source only once. */
     assert(file);   /* Non-NULL file object expected. */
 
     parser->read_handler = yaml_file_read_handler;
@@ -141,11 +176,11 @@ yaml_parser_set_input(yaml_parser_t *parser,
         yaml_read_handler_t *handler, void *data)
 {
     assert(parser); /* Non-NULL parser object expected. */
-    assert(!parser->reader); /* You can set the source only once. */
+    assert(!parser->read_handler);  /* You can set the source only once. */
     assert(handler);    /* Non-NULL read handler expected. */
 
     parser->read_handler = handler;
-    parser->read_handler_data = data
+    parser->read_handler_data = data;
 }
 
 /*
diff --git a/src/reader.c b/src/reader.c
index 787f785..440a88f 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -1,53 +1,304 @@
 
-#define RAW_BUFFER_SIZE 16384
-#define BUFFER_SIZE (RAW_BUFFER_SIZE*2) /* Should be enough for decoding 
-                                           the whole raw buffer. */
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
+
+#include <yaml/yaml.h>
+
+#include <assert.h>
+
+/* Check for the UTF-16-BE BOM. */
+#define IS_UTF16BE_BOM(pointer) ((pointer)[0] == 0xFE && (pointer)[1] == 0xFF)
+
+/* Check for the UTF-16-LE BOM. */
+#define IS_UTF16LE_BOM(pointer) ((pointer)[0] == 0xFF && (pointer)[1] == 0xFE)
+
+/* Get a UTF-16-BE character. */
+#define UTF16BE_CHAR(pointer)   ((pointer)[0] << 8 + (pointer)[1])
+
+/* Get a UTF-16-LE character. */
+#define UTF16LE_CHAR(pointer)   ((pointer)[0] + (pointer)[1] << 8)
+
+/*
+ * From http://www.ietf.org/rfc/rfc3629.txt:
+ *
+ *    Char. number range  |        UTF-8 octet sequence
+ *      (hexadecimal)    |              (binary)
+ *   --------------------+---------------------------------------------
+ *   0000 0000-0000 007F | 0xxxxxxx
+ *   0000 0080-0000 07FF | 110xxxxx 10xxxxxx
+ *   0000 0800-0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx
+ *   0001 0000-0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
+ */
+
+/* Get the length of a UTF-8 character (0 on error). */
+#define UTF8_LENGTH(pointer)    \
+    ((pointer)[0] < 0x80 ? 1 :  \
+    (pointer)[0] < 0xC0 ? 0 :   \
+    (pointer)[0] < 0xE0 ? 2 :   \
+    (pointer)[0] < 0xF0 ? 3 :   \
+    (pointer)[0] < 0xF8 ? 4 : 0)
+
+/* Get the value of the first byte of a UTF-8 sequence (0xFF on error). */
+#define UTF8_FIRST_CHUNK(pointer)   \
+    ((pointer)[0] < 0x80 ? (pointer)[0] & 0x7F :    \
+    (pointer)[0] < 0xC0 ? 0xFF :    \
+    (pointer)[0] < 0xE0 ? (pointer)[0] & 0x1F : \
+    (pointer)[0] < 0xF0 ? (pointer)[0] & 0x0F : \
+    (pointer)[0] < 0xF8 ? (pointer)[0] & 0x07 : 0xFF)
+
+/* Get the value of a non-first byte of a UTF-8 sequence (0xFF on error). */
+#define UTF8_NEXT_CHUNK(pointer)    \
+    ((pointer)[0] >= 0x80 && (pointer)[0] < 0xC0 ? (pointer)[0] & 0x3F : 0xFF)
+
+/* Determine the length of a UTF-8 character. */
 
 /*
  * Ensure that the buffer contains at least length characters.
  * Return 1 on success, 0 on failure.
+ *
+ * The length is supposed to be significantly less that the buffer size.
  */
 
 int
-yaml_parser_update_reader(yaml_parser_t *parser, size_t length)
+yaml_parser_update_buffer(yaml_parser_t *parser, size_t length)
 {
     /* If the EOF flag is set, do nothing. */
 
     if (parser->eof)
         return 1;
 
-    /* First, let us check that the buffers are allocated. */
+    /* Return if the buffer contains enough characters. */
+
+    if (parser->unread >= length)
+        return 1;
+
+    /* Determine the input encoding if it is not known yet. */
 
-    if (!parser->buffer) {
-        parser->buffer = yaml_malloc(BUFFER_SIZE);
-        if (!parser->buffer) {
-            parser->error = YAML_MEMORY_ERROR;
+    if (!parser->encoding) {
+        if (!yaml_parser_determine_encoding(parser))
             return 0;
-        }
-        parser->buffer_size = BUFFER_SIZE;
-        parser->buffer_pointer = parser->buffer;
-        parser->buffer_length = 0;
     }
 
-    if (!parser->raw_buffer) {
-        parser->raw_buffer = yaml_malloc(RAW_BUFFER_SIZE);
-        if (!parser->raw_buffer) {
-            parser->error = YAML_MEMORY_ERROR;
-            return 0;
+    /* Move the unread characters to the beginning of the buffer. */
+
+    if (parser->buffer < parser->pointer
+            && parser->pointer < parser->buffer_end) {
+        size_t size = parser->buffer_end - parser->pointer;
+        memmove(parser->buffer, parser->pointer, size);
+        parser->pointer = parser->buffer;
+        parser->buffer_end -= size;
+    }
+    else if (parser->pointer == parser->buffer_end) {
+        parser->pointer = parser->buffer;
+        parser->buffer_end = parser->buffer;
+    }
+
+    /* Fill the buffer until it has enough characters. */
+
+    while (parser->unread < length)
+    {
+        /* Fill the raw buffer. */
+
+        if (!yaml_parser_update_raw_buffer(parser)) return 0;
+
+        /* Decode the raw buffer. */
+
+        while (parser->raw_unread)
+        {
+            unsigned int ch;
+            int incomplete = 0;
+
+            /* Decode the next character. */
+
+            switch (parser->encoding)
+            {
+                case YAML_UTF8_ENCODING:
+
+                    unsigned int utf8_length = UTF8_LENGTH(parser->raw_pointer);
+                    unsigned int utf8_chunk;
+
+                    /* Check if the raw buffer contains an incomplete character. */
+
+                    if (utf8_length > parser->raw_unread) {
+                        if (parser->eof) {
+                            parser->error = YAML_READER_ERROR;
+                            return 0;
+                        }
+                        incomplete = 1;
+                    }
+
+                    /* Get the character checking it for validity. */
+
+                    utf8_chunk = UTF8_FIRST_CHUNK(parser->raw_pointer ++);
+                    if (utf8_chunk == 0xFF) {
+                        parser->error = YAML_READER_ERROR;
+                        return 0;
+                    }
+                    ch = utf8_chunk;
+                    parser->raw_unread --;
+                    while (-- utf8_length) {
+                        utf8_chunk = UTF8_NEXT_CHUNK(parser->raw_pointer ++);
+                        if (utf8_chunk == 0xFF) {
+                            parser->error = YAML_READER_ERROR;
+                            return 0;
+                        }
+                        ch = ch << 6 + utf8_chunk;
+                        parser->raw_unread --;
+                    }
+
+                    break;
+                
+                case YAML_UTF16LE_ENCODING:
+
+                    /* Check if the raw buffer contains an incomplete character. */
+
+                    if (parser->raw_unread < 2) {
+                        if (parser->eof) {
+                            parser->error = YAML_READER_ERROR;
+                            return 0;
+                        }
+                        incomplete = 1;
+                    }
+
+                    /* Get the current character. */
+
+                    ch = UTF16LE_CHAR(parser->raw_pointer);
+                    parser->raw_pointer += 2;
+                    parser->raw_unread -= 2;
+
+                    break;
+
+                case YAML_UTF16BE_ENCODING:
+
+                    /* Check if the raw buffer contains an incomplete character. */
+
+                    if (parser->raw_unread < 2) {
+                        if (parser->eof) {
+                            parser->error = YAML_READER_ERROR;
+                            return 0;
+                        }
+                        incomplete = 1;
+                    }
+
+                    /* Get the current character. */
+
+                    ch = UTF16BE_CHAR(parser->raw_pointer);
+                    parser->raw_pointer += 2;
+                    parser->raw_unread -= 2;
+
+                    break;
+            }
+
+            /*
+             * Check if the character is in the allowed range:
+             *      #x9 | #xA | #xD | [#x20-#x7E]               (8 bit)
+             *      | #x85 | [#xA0-#xD7FF] | [#xE000-#xFFFD]    (16 bit)
+             *      | [#x10000-#x10FFFF]                        (32 bit)
+             */
+
+            if (! (ch == 0x09 || ch == 0x0A || ch == 0x0D
+                        || (ch >= 0x20 && ch <= 0x7E)
+                        || (ch == 0x85) || (ch >= 0xA0 && ch <= 0xD7FF)
+                        || (ch >= 0xE000 && ch <= 0xFFFD)
+                        || (ch >= 0x10000 && ch <= 0x10FFFF))) {
+                parser->error = YAML_READER_ERROR;
+                return 0;
+            }
+
+            /* Finally put the character into the buffer. */
+
+            /* 0000 0000-0000 007F -> 0xxxxxxx */
+            if (ch <= 0x7F) {
+                *(parser->buffer_end++) = ch;
+            }
+            /* 0000 0080-0000 07FF -> 110xxxxx 10xxxxxx */
+            else if (ch <= 0x7FF) {
+                *(parser->buffer_end++) = 0xC0 + (ch >> 6) & 0x1F;
+                *(parser->buffer_end++) = 0x80 + ch & 0x3F;
+            }
+            /* 0000 0800-0000 FFFF -> 1110xxxx 10xxxxxx 10xxxxxx */
+            else if (ch <= 0xFFFF) {
+                *(parser->buffer_end++) = 0x80 + ch & 0x3F;
+                *(parser->buffer_end++) = 0xC0 + (ch >> 6) & 0x1F;
+                
+            }
+            /* 0001 0000-0010 FFFF -> 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
+            else {
+            }
         }
-        parser->raw_buffer_size = RAW_BUFFER_SIZE;
+        
     }
 
-    /* Next, determine the input encoding. */
+}
 
-    if (!parser->encoding) {
-        if (!yaml_parser_determine_encoding(parser))
+/*
+ * Determine the input stream encoding by checking the BOM symbol. If no BOM is
+ * found, the UTF-8 encoding is assumed. Return 1 on success, 0 on failure.
+ */
+
+int
+yaml_parser_determine_encoding(yaml_parser_t *parser)
+{
+    /* Ensure that we had enough bytes in the raw buffer. */
+
+    while (!parser->eof && parser->raw_unread < 2) {
+        if (!yaml_parser_update_raw_buffer(parser)) {
             return 0;
+        }
     }
 
-    /* more... */
+    /* Determine the encoding. */
 
+    if (parser->raw_unread >= 2 && IS_UTF16BE_BOM(parser->raw_pointer)) {
+        parser->encoding = YAML_UTF16BE_ENCODING;
+    }
+    else if (parser->raw_unread >= 2 && IS_UTF16LE_BOM(parser->raw_pointer)) {
+        parser->encoding = YAML_UTF16LE_ENCODING;
+    }
+    else {
+        parser->encoding = YAML_UTF8_ENCODING;
+    }
 }
 
+/*
+ * Update the raw buffer.
+ */
+
+int
+yaml_parser_update_raw_buffer(yaml_parser_t *parser)
+{
+    size_t size_read = 0;
+
+    /* Return if the raw buffer is full. */
+
+    if (parser->raw_unread == YAML_RAW_BUFFER_SIZE) return 1;
 
+    /* Return on EOF. */
+
+    if (parser->eof) return 1;
+
+    /* Move the remaining bytes in the raw buffer to the beginning. */
+
+    if (parser->raw_unread && parser->raw_buffer < parser->raw_pointer) {
+        memmove(parser->raw_buffer, parser->raw_pointer, parser->raw_unread);
+    }
+    parser->raw_pointer = parser->raw_buffer;
+
+    /* Call the read handler to fill the buffer. */
+
+    if (!parser->read_handler(parser->read_handler_data,
+                parser->raw_buffer + parser->raw_unread,
+                YAML_RAW_BUFFER_SIZE - parser->raw_unread,
+                &size_read)) {
+        parser->error = YAML_READER_ERROR;
+        return 0;
+    }
+    parser->raw_unread += size_read;
+    if (!size_read) {
+        parser->eof = 1;
+    }
+
+    return 1;
+}