ox 2.14.4 → 2.14.8

data/ext/ox/ox.h

@@ -16,9 +16,7 @@ extern "C" {
 #define RSTRING_NOT_MODIFIED
 
 #include "ruby.h"
-#if HAVE_RB_ENC_ASSOCIATE
 #include "ruby/encoding.h"
-#endif
 
 #if HAVE_RUBY_ST_H
 #include "ruby/st.h"
@@ -27,249 +25,227 @@ extern "C" {
 #include "st.h"
 #endif
 
-#include "cache.h"
-
-#include "err.h"
-#include "type.h"
 #include "attr.h"
+#include "err.h"
 #include "helper.h"
+#include "slotcache.h"
+#include "type.h"
 
 #define raise_error(msg, xml, current) _ox_raise_error(msg, xml, current, __FILE__, __LINE__)
 
-#define MAX_TEXT_LEN	4096
+#define MAX_TEXT_LEN 4096
 
-#define SILENT		0
-#define TRACE		1
-#define DEBUG		2
+#define SILENT 0
+#define TRACE 1
+#define DEBUG 2
 
-#define XSD_DATE	0x0001
-#define WITH_XML	0x0002
-#define WITH_INST	0x0004
-#define WITH_DTD	0x0008
-#define CIRCULAR	0x0010
+#define XSD_DATE 0x0001
+#define WITH_XML 0x0002
+#define WITH_INST 0x0004
+#define WITH_DTD 0x0008
+#define CIRCULAR 0x0010
 
-#define XSD_DATE_SET	0x0100
-#define WITH_XML_SET	0x0200
-#define WITH_INST_SET	0x0400
-#define WITH_DTD_SET	0x0800
-#define CIRCULAR_SET	0x1000
+#define XSD_DATE_SET 0x0100
+#define WITH_XML_SET 0x0200
+#define WITH_INST_SET 0x0400
+#define WITH_DTD_SET 0x0800
+#define CIRCULAR_SET 0x1000
 
 typedef enum {
-    UseObj	= 1,
-    UseAttr	= 2,
-    UseAttrSet	= 3,
-    UseArray	= 4,
-    UseAMember	= 5,
-    UseHash	= 6,
-    UseHashKey	= 7,
-    UseHashVal	= 8,
-    UseRange	= 9,
-    UseRangeAttr= 10,
-    UseRaw	= 11,
+    UseObj       = 1,
+    UseAttr      = 2,
+    UseAttrSet   = 3,
+    UseArray     = 4,
+    UseAMember   = 5,
+    UseHash      = 6,
+    UseHashKey   = 7,
+    UseHashVal   = 8,
+    UseRange     = 9,
+    UseRangeAttr = 10,
+    UseRaw       = 11,
 } Use;
 
 typedef enum {
-    StrictEffort    = 's',
-    TolerantEffort  = 't',
-    AutoEffort	    = 'a',
-    NoEffort	    = 0,
+    StrictEffort   = 's',
+    TolerantEffort = 't',
+    AutoEffort     = 'a',
+    NoEffort       = 0,
 } Effort;
 
-typedef enum {
-    Yes	   = 'y',
-    No	   = 'n',
-    NotSet = 0
-} YesNo;
+typedef enum { Yes = 'y', No = 'n', NotSet = 0 } YesNo;
 
-typedef enum {
-    ObjMode        = 'o',
-    GenMode        = 'g',
-    LimMode        = 'l',
-    HashMode       = 'h',
-    HashNoAttrMode = 'n',
-    NoMode         = 0
-} LoadMode;
+typedef enum { ObjMode = 'o', GenMode = 'g', LimMode = 'l', HashMode = 'h', HashNoAttrMode = 'n', NoMode = 0 } LoadMode;
 
 typedef enum {
-    OffSkip  = 'o',
-    NoSkip   = 'n',
-    CrSkip   = 'r',
-    SpcSkip  = 's',
+    OffSkip = 'o',
+    NoSkip  = 'n',
+    CrSkip  = 'r',
+    SpcSkip = 's',
 } SkipMode;
 
-typedef struct _pInfo	*PInfo;
+typedef struct _pInfo *PInfo;
 
 typedef struct _parseCallbacks {
-    void	(*instruct)(PInfo pi, const char *target, Attr attrs, const char *content);
-    void	(*add_doctype)(PInfo pi, const char *docType);
-    void	(*add_comment)(PInfo pi, const char *comment);
-    void	(*add_cdata)(PInfo pi, const char *cdata, size_t len);
-    void	(*add_text)(PInfo pi, char *text, int closed);
-    void	(*add_element)(PInfo pi, const char *ename, Attr attrs, int hasChildren);
-    void	(*end_element)(PInfo pi, const char *ename);
-    void	(*finish)(PInfo pi);
-} *ParseCallbacks;
+    void (*instruct)(PInfo pi, const char *target, Attr attrs, const char *content);
+    void (*add_doctype)(PInfo pi, const char *docType);
+    void (*add_comment)(PInfo pi, const char *comment);
+    void (*add_cdata)(PInfo pi, const char *cdata, size_t len);
+    void (*add_text)(PInfo pi, char *text, int closed);
+    void (*add_element)(PInfo pi, const char *ename, Attr attrs, int hasChildren);
+    void (*end_element)(PInfo pi, const char *ename);
+    void (*finish)(PInfo pi);
+} * ParseCallbacks;
 
 typedef struct _circArray {
-    VALUE		obj_array[1024];
-    VALUE		*objs;
-    unsigned long	size; /* allocated size or initial array size */
-    unsigned long	cnt;
-} *CircArray;
+    VALUE         obj_array[1024];
+    VALUE        *objs;
+    unsigned long size; /* allocated size or initial array size */
+    unsigned long cnt;
+} * CircArray;
 
 typedef struct _options {
-    char		encoding[64];	// encoding, stored in the option to avoid GC invalidation in default values
-    char		margin[128];	// left margin for dumping
-    int			indent;		// indention for dump, default 2
-    int			trace;		// trace level
-    char		margin_len;	// margin length
-    char		with_dtd;	// YesNo
-    char		with_xml;	// YesNo
-    char		with_instruct;	// YesNo
-    char		circular;	// YesNo
-    char		xsd_date;	// YesNo
-    char		mode;		// LoadMode
-    char		effort;		// Effort
-    char		sym_keys;	// symbolize keys
-    char		skip;		// skip mode
-    char		smart;		// YesNo sax smart mode
-    char		convert_special;// boolean true or false
-    char		allow_invalid;	// YesNo
-    char		no_empty;	// boolean - no empty elements when dumping
-    char		with_cdata;	// boolean - hash_load should include cdata
-    char		inv_repl[12];	// max 10 valid characters, first character is the length
-    char		strip_ns[64];	// namespace to strip, \0 is no-strip, \* is all, else only matches
-    struct _hints	*html_hints;	// html hints
-    VALUE		attr_key_mod;
-    VALUE		element_key_mod;
-#if HAVE_RB_ENC_ASSOCIATE
-    rb_encoding		*rb_enc;
-#else
-    void		*rb_enc;
-#endif
-} *Options;
+    char           encoding[64];     // encoding, stored in the option to avoid GC invalidation in default values
+    char           margin[128];      // left margin for dumping
+    int            indent;           // indention for dump, default 2
+    int            trace;            // trace level
+    char           margin_len;       // margin length
+    char           with_dtd;         // YesNo
+    char           with_xml;         // YesNo
+    char           with_instruct;    // YesNo
+    char           circular;         // YesNo
+    char           xsd_date;         // YesNo
+    char           mode;             // LoadMode
+    char           effort;           // Effort
+    char           sym_keys;         // symbolize keys
+    char           skip;             // skip mode
+    char           smart;            // YesNo sax smart mode
+    char           convert_special;  // boolean true or false
+    char           allow_invalid;    // YesNo
+    char           no_empty;         // boolean - no empty elements when dumping
+    char           with_cdata;       // boolean - hash_load should include cdata
+    char           inv_repl[12];     // max 10 valid characters, first character is the length
+    char           strip_ns[64];     // namespace to strip, \0 is no-strip, \* is all, else only matches
+    struct _hints *html_hints;       // html hints
+    VALUE          attr_key_mod;
+    VALUE          element_key_mod;
+    rb_encoding *rb_enc;
+} * Options;
 
 // parse information structure
 struct _pInfo {
-    struct _helperStack	helpers;
-    struct _err		err;
-    char		*str;		// buffer being read from
-    char		*end;		// end of original string
-    char		*s;		// current position in buffer
-    VALUE		obj;
-    ParseCallbacks	pcb;
-    CircArray		circ_array;
-    unsigned long	id;		// set for text types when cirs_array is set
-    Options		options;
-    VALUE		*marked;
-    int			mark_size;	// allocated size
-    int			mark_cnt;
-    char		last;		// last character read, rarely set
+    struct _helperStack helpers;
+    struct _err         err;
+    char               *str;  // buffer being read from
+    char               *end;  // end of original string
+    char               *s;    // current position in buffer
+    VALUE               obj;
+    ParseCallbacks      pcb;
+    CircArray           circ_array;
+    unsigned long       id;  // set for text types when cirs_array is set
+    Options             options;
+    VALUE              *marked;
+    int                 mark_size;  // allocated size
+    int                 mark_cnt;
+    char                last;  // last character read, rarely set
 };
 
-extern VALUE	ox_parse(char *xml, size_t len, ParseCallbacks pcb, char **endp, Options options, Err err);
-extern void	_ox_raise_error(const char *msg, const char *xml, const char *current, const char* file, int line);
-
-extern void	ox_sax_define(void);
-
-extern char*	ox_write_obj_to_str(VALUE obj, Options copts);
-extern void	ox_write_obj_to_file(VALUE obj, const char *path, Options copts);
-
-extern struct _options	ox_default_options;
-
-extern VALUE	Ox;
-
-extern ID	ox_abort_id;
-extern ID	ox_at_column_id;
-extern ID	ox_at_content_id;
-extern ID	ox_at_id;
-extern ID	ox_at_line_id;
-extern ID	ox_at_pos_id;
-extern ID	ox_at_value_id;
-extern ID	ox_attr_id;
-extern ID	ox_attr_value_id;
-extern ID	ox_attrs_done_id;
-extern ID	ox_attributes_id;
-extern ID	ox_beg_id;
-extern ID	ox_bigdecimal_id;
-extern ID	ox_call_id;
-extern ID	ox_cdata_id;
-extern ID	ox_comment_id;
-extern ID	ox_den_id;
-extern ID	ox_doctype_id;
-extern ID	ox_end_element_id;
-extern ID	ox_end_id;
-extern ID	ox_end_instruct_id;
-extern ID	ox_error_id;
-extern ID	ox_excl_id;
-extern ID	ox_external_encoding_id;
-extern ID	ox_fileno_id;
-extern ID	ox_force_encoding_id;
-extern ID	ox_inspect_id;
-extern ID	ox_instruct_id;
-extern ID	ox_jd_id;
-extern ID	ox_keys_id;
-extern ID	ox_local_id;
-extern ID	ox_mesg_id;
-extern ID	ox_message_id;
-extern ID	ox_new_id;
-extern ID	ox_nodes_id;
-extern ID	ox_num_id;
-extern ID	ox_parse_id;
-extern ID	ox_pos_id;
-extern ID	ox_read_id;
-extern ID	ox_readpartial_id;
-extern ID	ox_start_element_id;
-extern ID	ox_string_id;
-extern ID	ox_text_id;
-extern ID	ox_to_c_id;
-extern ID	ox_to_s_id;
-extern ID	ox_to_sym_id;
-extern ID	ox_tv_sec_id;
-extern ID	ox_tv_nsec_id;
-extern ID	ox_tv_usec_id;
-extern ID	ox_value_id;
-
-#if HAVE_RB_ENC_ASSOCIATE
-extern rb_encoding	*ox_utf8_encoding;
-#else
-extern void		*ox_utf8_encoding;
-#endif
-
-extern VALUE	ox_empty_string;
-extern VALUE	ox_encoding_sym;
-extern VALUE	ox_indent_sym;
-extern VALUE	ox_size_sym;
-extern VALUE	ox_standalone_sym;
-extern VALUE	ox_sym_bank; // Array
-extern VALUE	ox_version_sym;
-extern VALUE	ox_zero_fixnum;
-
-extern VALUE	ox_bigdecimal_class;
-extern VALUE	ox_date_class;
-extern VALUE	ox_stringio_class;
-extern VALUE	ox_struct_class;
-extern VALUE	ox_time_class;
-
-extern VALUE	ox_document_clas;
-extern VALUE	ox_element_clas;
-extern VALUE	ox_instruct_clas;
-extern VALUE	ox_bag_clas;
-extern VALUE	ox_comment_clas;
-extern VALUE	ox_raw_clas;
-extern VALUE	ox_doctype_clas;
-extern VALUE	ox_cdata_clas;
-
-extern Cache	ox_symbol_cache;
-extern Cache	ox_class_cache;
-extern Cache	ox_attr_cache;
-
-extern void	ox_init_builder(VALUE ox);
+extern VALUE ox_parse(char *xml, size_t len, ParseCallbacks pcb, char **endp, Options options, Err err);
+extern void  _ox_raise_error(const char *msg, const char *xml, const char *current, const char *file, int line);
+
+extern void ox_sax_define(void);
+
+extern char *ox_write_obj_to_str(VALUE obj, Options copts);
+extern void  ox_write_obj_to_file(VALUE obj, const char *path, Options copts);
+
+extern struct _options ox_default_options;
+
+extern VALUE Ox;
+
+extern ID ox_abort_id;
+extern ID ox_at_column_id;
+extern ID ox_at_content_id;
+extern ID ox_at_id;
+extern ID ox_at_line_id;
+extern ID ox_at_pos_id;
+extern ID ox_at_value_id;
+extern ID ox_attr_id;
+extern ID ox_attr_value_id;
+extern ID ox_attrs_done_id;
+extern ID ox_attributes_id;
+extern ID ox_beg_id;
+extern ID ox_bigdecimal_id;
+extern ID ox_call_id;
+extern ID ox_cdata_id;
+extern ID ox_comment_id;
+extern ID ox_den_id;
+extern ID ox_doctype_id;
+extern ID ox_end_element_id;
+extern ID ox_end_id;
+extern ID ox_end_instruct_id;
+extern ID ox_error_id;
+extern ID ox_excl_id;
+extern ID ox_external_encoding_id;
+extern ID ox_fileno_id;
+extern ID ox_force_encoding_id;
+extern ID ox_inspect_id;
+extern ID ox_instruct_id;
+extern ID ox_jd_id;
+extern ID ox_keys_id;
+extern ID ox_local_id;
+extern ID ox_mesg_id;
+extern ID ox_message_id;
+extern ID ox_new_id;
+extern ID ox_nodes_id;
+extern ID ox_num_id;
+extern ID ox_parse_id;
+extern ID ox_pos_id;
+extern ID ox_read_id;
+extern ID ox_readpartial_id;
+extern ID ox_start_element_id;
+extern ID ox_string_id;
+extern ID ox_text_id;
+extern ID ox_to_c_id;
+extern ID ox_to_s_id;
+extern ID ox_to_sym_id;
+extern ID ox_tv_sec_id;
+extern ID ox_tv_nsec_id;
+extern ID ox_tv_usec_id;
+extern ID ox_value_id;
+
+extern rb_encoding *ox_utf8_encoding;
+
+extern VALUE ox_empty_string;
+extern VALUE ox_encoding_sym;
+extern VALUE ox_indent_sym;
+extern VALUE ox_size_sym;
+extern VALUE ox_standalone_sym;
+extern VALUE ox_sym_bank;  // Array
+extern VALUE ox_version_sym;
+extern VALUE ox_zero_fixnum;
+
+extern VALUE ox_bigdecimal_class;
+extern VALUE ox_date_class;
+extern VALUE ox_stringio_class;
+extern VALUE ox_struct_class;
+extern VALUE ox_time_class;
+
+extern VALUE ox_document_clas;
+extern VALUE ox_element_clas;
+extern VALUE ox_instruct_clas;
+extern VALUE ox_bag_clas;
+extern VALUE ox_comment_clas;
+extern VALUE ox_raw_clas;
+extern VALUE ox_doctype_clas;
+extern VALUE ox_cdata_clas;
+
+extern SlotCache ox_class_cache;
+
+extern void ox_init_builder(VALUE ox);
 
 #if defined(__cplusplus)
 #if 0
 { /* satisfy cc-mode */
 #endif
-}  /* extern "C" { */
+} /* extern "C" { */
 #endif
 #endif /* OX_H */

data/ext/ox/parse.c

@@ -5,13 +5,16 @@
 
 #include <stdlib.h>
 #include <errno.h>
+#include <stdbool.h>
 #include <stdio.h>
 #include <string.h>
+#include <strings.h>
 
 #include "ruby.h"
 #include "ox.h"
 #include "err.h"
 #include "attr.h"
+#include "intern.h"
 #include "helper.h"
 #include "special.h"
 
@@ -93,6 +96,30 @@ next_white(PInfo pi) {
     }
 }
 
+static void fix_newlines(char *buf) {
+#if HAVE_INDEX
+    if (NULL != index(buf, '\r')) {
+#endif
+	char	*s = buf;
+	char	*d = buf;
+
+	for (; '\0' != *s; s++) {
+	    if ('\r' == *s) {
+		if ('\n' == *(s + 1)) {
+		    continue;
+		}
+		*s = '\n';
+	    } else if (d < s) {
+		*d = *s;
+	    }
+	    d++;
+	}
+	*d = '\0';
+#if HAVE_INDEX
+    }
+#endif
+}
+
 static void
 mark_pi_cb(void *ptr) {
     if (NULL != ptr) {
@@ -211,31 +238,11 @@ ox_parse(char *xml, size_t len, ParseCallbacks pcb, char **endp, Options options
     return pi.obj;
 }
 
-static char*
-gather_content(const char *src, char *content, size_t len) {
-    for (; 0 < len; src++, content++, len--) {
-	switch (*src) {
-	case '?':
-	    if ('>' == *(src + 1)) {
-		*content = '\0';
-		return (char*)(src + 1);
-	    }
-	    *content = *src;
-	    break;
-	case '\0':
-	    return 0;
-	default:
-	    *content = *src;
-	    break;
-	}
-    }
-    return 0;
-}
-
 // Entered after the "<?" sequence. Ready to read the rest.
 static void
 read_instruction(PInfo pi) {
-    char		content[1024];
+    char		content[256];
+    char		*content_ptr;
     struct _attrStack	attrs;
     char		*attr_name;
     char		*attr_value;
@@ -243,7 +250,8 @@ read_instruction(PInfo pi) {
     char		*end;
     char		c;
     char		*cend;
-    int			attrs_ok = 1;
+    size_t		size;
+    bool		attrs_ok = true;
 
     *content = '\0';
     attr_stack_init(&attrs);
@@ -251,10 +259,33 @@ read_instruction(PInfo pi) {
 	return;
     }
     end = pi->s;
-    if (0 == (cend = gather_content(pi->s, content, sizeof(content) - 1))) {
-	set_error(&pi->err, "processing instruction content too large or not terminated", pi->str, pi->s);
-	return;
+    for (; true; pi->s++) {
+        switch (*pi->s) {
+        case '?':
+            if ('>' == *(pi->s + 1)) {
+		pi->s++;
+		goto DONE;
+            }
+            break;
+        case '\0':
+	    set_error(&pi->err, "processing instruction not terminated", pi->str, pi->s);
+	    return;
+        default:
+	    break;
+        }
     }
+DONE:
+    cend = pi->s;
+    size = cend - end - 1;
+    pi->s = end;
+    if (size < sizeof(content)) {
+	content_ptr = content;
+    } else {
+	content_ptr = ALLOC_N(char, size + 1);
+    }
+    memcpy(content_ptr, end, size);
+    content_ptr[size] = '\0';
+
     next_non_white(pi);
     c = *pi->s;
     *end = '\0'; // terminate name
@@ -274,7 +305,7 @@ read_instruction(PInfo pi) {
 	    end = pi->s;
 	    next_non_white(pi);
 	    if ('=' != *pi->s++) {
-		attrs_ok = 0;
+		attrs_ok = false;
 		break;
 	    }
 	    *end = '\0'; // terminate name
@@ -311,10 +342,13 @@ read_instruction(PInfo pi) {
 	if (attrs_ok) {
 	    pi->pcb->instruct(pi, target, attrs.head, 0);
 	} else {
-	    pi->pcb->instruct(pi, target, attrs.head, content);
+	    pi->pcb->instruct(pi, target, attrs.head, content_ptr);
 	}
     }
     attr_stack_cleanup(&attrs);
+    if (content_ptr != content) {
+	xfree(content_ptr);
+    }
 }
 
 static void
@@ -361,10 +395,10 @@ read_delimited(PInfo pi, char end) {
 // that. Ready to read the rest.
 static void
 read_doctype(PInfo pi) {
-    char	*docType;
+    char	*doctype;
 
     next_non_white(pi);
-    docType = pi->s;
+    doctype = pi->s;
     read_delimited(pi, '>');
     if (err_has(&pi->err)) {
 	return;
@@ -373,7 +407,8 @@ read_doctype(PInfo pi) {
     *pi->s = '\0';
     pi->s++;
     if (0 != pi->pcb->add_doctype) {
-	pi->pcb->add_doctype(pi, docType);
+	fix_newlines(doctype);
+	pi->pcb->add_doctype(pi, doctype);
     }
 }
 
@@ -409,6 +444,7 @@ read_comment(PInfo pi) {
     *end = '\0'; // in case the comment was blank
     pi->s = end + 3;
     if (0 != pi->pcb->add_comment) {
+	fix_newlines(comment);
 	pi->pcb->add_comment(pi, comment);
     }
 }
@@ -769,9 +805,11 @@ read_text(PInfo pi) {
     }
     *b = '\0';
     if (0 != alloc_buf) {
+	fix_newlines(alloc_buf);
 	pi->pcb->add_text(pi, alloc_buf, ('/' == *(pi->s + 1)));
 	xfree(alloc_buf);
     } else {
+	fix_newlines(buf);
 	pi->pcb->add_text(pi, buf, ('/' == *(pi->s + 1)));
     }
 }
@@ -838,9 +876,11 @@ read_reduced_text(PInfo pi) {
     }
     *b = '\0';
     if (0 != alloc_buf) {
+	fix_newlines(alloc_buf);
 	pi->pcb->add_text(pi, alloc_buf, ('/' == *(pi->s + 1)));
 	xfree(alloc_buf);
     } else {
+	fix_newlines(buf);
 	pi->pcb->add_text(pi, buf, ('/' == *(pi->s + 1)));
     }
 }
@@ -899,6 +939,7 @@ read_cdata(PInfo pi) {
     *end = '\0';
     pi->s = end + 3;
     if (0 != pi->pcb->add_cdata) {
+	fix_newlines(start);
 	pi->pcb->add_cdata(pi, start, end - start);
     }
 }