blob: b89a28303d05a350eb87059034b62b3c7fe48d09 [file] [log] [blame]
Radek Krejcid91dbaf2018-09-21 15:51:39 +02001/**
2 * @file xml.h
3 * @author Radek Krejci <rkrejci@cesnet.cz>
4 * @brief Generic XML parser routines.
5 *
6 * Copyright (c) 2015 - 2018 CESNET, z.s.p.o.
7 *
8 * This source code is licensed under BSD 3-Clause License (the "License").
9 * You may not use this file except in compliance with the License.
10 * You may obtain a copy of the License at
11 *
12 * https://opensource.org/licenses/BSD-3-Clause
13 */
14
15#ifndef LY_XML_H_
16#define LY_XML_H_
17
Radek Krejcie7b95092019-05-15 11:03:07 +020018#include <stddef.h>
Radek Krejcid91dbaf2018-09-21 15:51:39 +020019#include <stdint.h>
20
Radek Krejcie7b95092019-05-15 11:03:07 +020021#include "log.h"
Radek Krejcid91dbaf2018-09-21 15:51:39 +020022#include "set.h"
23
Radek Krejci535ea9f2020-05-29 16:01:05 +020024struct ly_ctx;
Radek Krejci241f6b52020-05-21 18:13:49 +020025struct ly_out;
Michal Vasko52927e22020-03-16 17:26:14 +010026struct ly_prefix;
Radek Krejcie7b95092019-05-15 11:03:07 +020027
Radek Krejcib1646a92018-11-02 16:08:26 +010028/* Macro to test if character is whitespace */
29#define is_xmlws(c) (c == 0x20 || c == 0x9 || c == 0xa || c == 0xd)
30
31/* Macro to test if character is allowed to be a first character of an qualified identifier */
32#define is_xmlqnamestartchar(c) ((c >= 'a' && c <= 'z') || c == '_' || \
33 (c >= 'A' && c <= 'Z') || /* c == ':' || */ \
34 (c >= 0x370 && c <= 0x1fff && c != 0x37e ) || \
35 (c >= 0xc0 && c <= 0x2ff && c != 0xd7 && c != 0xf7) || c == 0x200c || \
36 c == 0x200d || (c >= 0x2070 && c <= 0x218f) || \
37 (c >= 0x2c00 && c <= 0x2fef) || (c >= 0x3001 && c <= 0xd7ff) || \
38 (c >= 0xf900 && c <= 0xfdcf) || (c >= 0xfdf0 && c <= 0xfffd) || \
39 (c >= 0x10000 && c <= 0xeffff))
40
41/* Macro to test if character is allowed to be used in an qualified identifier */
42#define is_xmlqnamechar(c) ((c >= 'a' && c <= 'z') || c == '_' || c == '-' || \
43 (c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || /* c == ':' || */ \
44 c == '.' || c == 0xb7 || (c >= 0x370 && c <= 0x1fff && c != 0x37e ) ||\
45 (c >= 0xc0 && c <= 0x2ff && c != 0xd7 && c != 0xf7) || c == 0x200c || \
46 c == 0x200d || (c >= 0x300 && c <= 0x36f) || \
47 (c >= 0x2070 && c <= 0x218f) || (c >= 0x2030f && c <= 0x2040) || \
48 (c >= 0x2c00 && c <= 0x2fef) || (c >= 0x3001 && c <= 0xd7ff) || \
49 (c >= 0xf900 && c <= 0xfdcf) || (c >= 0xfdf0 && c <= 0xfffd) || \
50 (c >= 0x10000 && c <= 0xeffff))
51
Radek Krejcid91dbaf2018-09-21 15:51:39 +020052struct lyxml_ns {
Radek Krejci4b74d5e2018-09-26 14:30:55 +020053 char *prefix; /* prefix of the namespace, NULL for the default namespace */
54 char *uri; /* namespace URI */
Michal Vaskob36053d2020-03-26 15:49:30 +010055 uint32_t depth; /* depth level of the element to maintain the list of accessible namespace definitions */
Radek Krejcid91dbaf2018-09-21 15:51:39 +020056};
57
Radek Krejcib1890642018-10-03 14:05:40 +020058/* element tag identifier for matching opening and closing tags */
59struct lyxml_elem {
60 const char *prefix;
61 const char *name;
62 size_t prefix_len;
63 size_t name_len;
64};
65
Radek Krejci28e8cb52019-03-08 11:31:31 +010066/**
67 * @brief Status of the parser providing information what is expected next (which function is supposed to be called).
68 */
Radek Krejcid91dbaf2018-09-21 15:51:39 +020069enum LYXML_PARSER_STATUS {
Michal Vaskob36053d2020-03-26 15:49:30 +010070 LYXML_ELEMENT, /* opening XML element parsed */
71 LYXML_ELEM_CLOSE, /* closing XML element parsed */
72 LYXML_ELEM_CONTENT, /* XML element context parsed */
73 LYXML_ATTRIBUTE, /* XML attribute parsed */
74 LYXML_ATTR_CONTENT, /* XML attribute content parsed */
Radek Krejcib1890642018-10-03 14:05:40 +020075 LYXML_END /* end of input data */
Radek Krejcid91dbaf2018-09-21 15:51:39 +020076};
77
Michal Vaskob36053d2020-03-26 15:49:30 +010078struct lyxml_ctx {
79 enum LYXML_PARSER_STATUS status; /* status providing information about the last parsed object, following attributes
80 are filled based on it */
81 union {
Michal Vasko8cef5232020-06-15 17:59:47 +020082 const char *prefix; /* LYXML_ELEMENT, LYXML_ATTRIBUTE - elem/attr prefix */
83 const char *value; /* LYXML_ELEM_CONTENT, LYXML_ATTR_CONTENT - elem/attr value */
Michal Vaskob36053d2020-03-26 15:49:30 +010084 };
85 union {
Michal Vasko8cef5232020-06-15 17:59:47 +020086 size_t prefix_len; /* LYXML_ELEMENT, LYXML_ATTRIBUTE - elem/attr prefix length */
87 size_t value_len; /* LYXML_ELEM_CONTENT, LYXML_ATTR_CONTENT - elem/attr value length */
Michal Vaskob36053d2020-03-26 15:49:30 +010088 };
89 union {
Michal Vasko8cef5232020-06-15 17:59:47 +020090 const char *name; /* LYXML_ELEMENT, LYXML_ATTRIBUTE - elem/attr name */
91 int ws_only; /* LYXML_ELEM_CONTENT, LYXML_ATTR_CONTENT - whether elem/attr value is empty/white-space only */
Michal Vaskob36053d2020-03-26 15:49:30 +010092 };
93 union {
Michal Vasko8cef5232020-06-15 17:59:47 +020094 size_t name_len; /* LYXML_ELEMENT, LYXML_ATTRIBUTE - elem/attr name length */
95 int dynamic; /* LYXML_ELEM_CONTENT, LYXML_ATTR_CONTENT - whether elem/attr value is dynamically allocated */
Michal Vaskob36053d2020-03-26 15:49:30 +010096 };
97
98 const struct ly_ctx *ctx;
Michal Vasko8cef5232020-06-15 17:59:47 +020099 uint64_t line; /* current line */
Michal Vasko63f3d842020-07-08 10:10:14 +0200100 struct ly_in *in; /* input structure */
Radek Krejcib1890642018-10-03 14:05:40 +0200101 struct ly_set elements; /* list of not-yet-closed elements */
Michal Vaskob36053d2020-03-26 15:49:30 +0100102 struct ly_set ns; /* handled with LY_SET_OPT_USEASLIST */
Radek Krejcid91dbaf2018-09-21 15:51:39 +0200103};
104
Michal Vasko8cef5232020-06-15 17:59:47 +0200105/**
106 * @brief Create a new XML parser context and start parsing.
107 *
108 * @param[in] ctx libyang context.
Michal Vasko63f3d842020-07-08 10:10:14 +0200109 * @param[in] in Input structure.
Michal Vasko8cef5232020-06-15 17:59:47 +0200110 * @param[out] xmlctx New XML context with status ::LYXML_ELEMENT.
111 * @return LY_ERR value.
112 */
Michal Vasko63f3d842020-07-08 10:10:14 +0200113LY_ERR lyxml_ctx_new(const struct ly_ctx *ctx, struct ly_in *in, struct lyxml_ctx **xmlctx);
Radek Krejci4b74d5e2018-09-26 14:30:55 +0200114
Michal Vasko8cef5232020-06-15 17:59:47 +0200115/**
116 * @brief Move to the next XML artefact and update parser status.
117 *
118 * LYXML_ELEMENT (-> LYXML_ATTRIBUTE -> LYXML_ATTR_CONTENT)* -> LYXML_ELEM_CONTENT -> LYXML_ELEM_CLOSE ...
119 * -> LYXML_ELEMENT ...
120 *
121 * @param[in] xmlctx XML context to move.
122 * @return LY_ERR value.
123 */
Michal Vaskob36053d2020-03-26 15:49:30 +0100124LY_ERR lyxml_ctx_next(struct lyxml_ctx *xmlctx);
Michal Vasko52927e22020-03-16 17:26:14 +0100125
Michal Vasko8cef5232020-06-15 17:59:47 +0200126/**
127 * @brief Peek at the next XML parser status without changing it.
128 *
129 * @param[in] xmlctx XML context to use.
130 * @param[out] next Next XML parser status.
131 * @return LY_ERR value.
132 */
Michal Vaskob36053d2020-03-26 15:49:30 +0100133LY_ERR lyxml_ctx_peek(struct lyxml_ctx *xmlctx, enum LYXML_PARSER_STATUS *next);
Radek Krejci4b74d5e2018-09-26 14:30:55 +0200134
135/**
Radek Krejci4b74d5e2018-09-26 14:30:55 +0200136 * @brief Get a namespace record for the given prefix in the current context.
137 *
Michal Vaskob36053d2020-03-26 15:49:30 +0100138 * @param[in] xmlctx XML context to work with.
Radek Krejci4b74d5e2018-09-26 14:30:55 +0200139 * @param[in] prefix Pointer to the namespace prefix as taken from lyxml_get_attribute() or lyxml_get_element().
140 * Can be NULL for default namespace.
141 * @param[in] prefix_len Length of the prefix string (since it is not NULL-terminated when returned from lyxml_get_attribute() or
142 * lyxml_get_element()).
143 * @return The namespace record or NULL if the record for the specified prefix not found.
144 */
Michal Vaskob36053d2020-03-26 15:49:30 +0100145const struct lyxml_ns *lyxml_ns_get(struct lyxml_ctx *xmlctx, const char *prefix, size_t prefix_len);
Radek Krejci4b74d5e2018-09-26 14:30:55 +0200146
147/**
Radek Krejcie7b95092019-05-15 11:03:07 +0200148 * @brief Print the given @p text as XML string which replaces some of the characters which cannot appear in XML data.
149 *
150 * @param[in] out Output structure for printing.
151 * @param[in] text String to print.
152 * @param[in] attribute Flag for attribute's value where a double quotes must be replaced.
153 * @return LY_ERR values.
154 */
Radek Krejci241f6b52020-05-21 18:13:49 +0200155LY_ERR lyxml_dump_text(struct ly_out *out, const char *text, int attribute);
Radek Krejcie7b95092019-05-15 11:03:07 +0200156
157/**
Radek Krejcib1890642018-10-03 14:05:40 +0200158 * @brief Remove the allocated working memory of the context.
159 *
Michal Vaskob36053d2020-03-26 15:49:30 +0100160 * @param[in] xmlctx XML context to clear.
Radek Krejcib1890642018-10-03 14:05:40 +0200161 */
Michal Vaskob36053d2020-03-26 15:49:30 +0100162void lyxml_ctx_free(struct lyxml_ctx *xmlctx);
Radek Krejcib1890642018-10-03 14:05:40 +0200163
Michal Vasko52927e22020-03-16 17:26:14 +0100164/**
165 * @brief Find all possible prefixes in a value.
166 *
Michal Vaskob36053d2020-03-26 15:49:30 +0100167 * @param[in] xmlctx XML context to use.
Michal Vasko52927e22020-03-16 17:26:14 +0100168 * @param[in] value Value to check.
169 * @param[in] value_len Value length.
170 * @param[out] val_prefs Array of found prefixes.
171 * @return LY_ERR value.
172 */
Michal Vaskob36053d2020-03-26 15:49:30 +0100173LY_ERR lyxml_get_prefixes(struct lyxml_ctx *xmlctx, const char *value, size_t value_len, struct ly_prefix **val_prefs);
Michal Vasko52927e22020-03-16 17:26:14 +0100174
175/**
176 * @brief Compare values and their prefix mappings.
177 *
178 * @param[in] value1 First value.
179 * @param[in] prefs1 First value prefixes.
180 * @param[in] value2 Second value.
181 * @param[in] prefs2 Second value prefixes.
182 * @return LY_SUCCESS if values are equal.
183 * @return LY_ENOT if values are not equal.
184 * @return LY_ERR on error.
185 */
186LY_ERR lyxml_value_compare(const char *value1, const struct ly_prefix *prefs1, const char *value2,
187 const struct ly_prefix *prefs2);
188
Radek Krejcib416be62018-10-01 14:51:45 +0200189#endif /* LY_XML_H_ */