lotobo/regex-tools
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 6 Wiki Activity

feat: publish Regex Tools 0.2.0

Browse Source
This commit is contained in:
Albrecht Degering
2026-07-27 01:06:57 +02:00
parent 873b9b218d
commit 4951c966d4
180 changed files with 35344 additions and 503 deletions
Download Patch File Download Diff File Expand all files Collapse all files

236
engines/pcre2/pcre2_bridge.h Normal file
View File

@@ -0,0 +1,236 @@
/* SPDX-License-Identifier: GPL-3.0-or-later */
#ifndef REGEX_TOOLS_PCRE2_BRIDGE_H
#define REGEX_TOOLS_PCRE2_BRIDGE_H
#include <stdint.h>
#define REGEX_PCRE2_BRIDGE_ABI_VERSION 3u
#define REGEX_PCRE2_MAX_PATTERN_BYTES 1048576u
#define REGEX_PCRE2_MAX_SUBJECT_BYTES 16777216u
#define REGEX_PCRE2_MAX_REPLACEMENT_BYTES 262144u
#define REGEX_PCRE2_MAX_MATCHES 10000u
#define REGEX_PCRE2_MAX_CAPTURE_ROWS 100000u
#define REGEX_PCRE2_MAX_CAPTURE_GROUPS 1000u
#define REGEX_PCRE2_MAX_MATCH_LIMIT 100000000u
#define REGEX_PCRE2_MAX_DEPTH_LIMIT 100000u
#define REGEX_PCRE2_MAX_HEAP_LIMIT_KIB 131072u
#define REGEX_PCRE2_MAX_TRACE_EVENTS 50000u
#define REGEX_PCRE2_MAX_TRACE_BYTES 10485760u
#define REGEX_PCRE2_MAX_TRACE_MARK_BYTES 1024u
#define REGEX_PCRE2_UNSET_OFFSET UINT32_MAX
enum regex_pcre2_flag {
REGEX_PCRE2_CASELESS = 1u << 0,
REGEX_PCRE2_MULTILINE = 1u << 1,
REGEX_PCRE2_DOTALL = 1u << 2,
REGEX_PCRE2_EXTENDED = 1u << 3,
REGEX_PCRE2_UNGREEDY = 1u << 4,
REGEX_PCRE2_UTF = 1u << 5,
REGEX_PCRE2_UCP = 1u << 6,
REGEX_PCRE2_DUPNAMES = 1u << 7,
REGEX_PCRE2_GLOBAL = 1u << 8
};
enum regex_pcre2_config_flag {
REGEX_PCRE2_CONFIG_UNICODE = 1u << 0,
REGEX_PCRE2_CONFIG_JIT = 1u << 1
};
enum regex_pcre2_bridge_error {
REGEX_PCRE2_ERROR_INVALID_ARGUMENT = -10001,
REGEX_PCRE2_ERROR_UNSUPPORTED_FLAGS = -10002,
REGEX_PCRE2_ERROR_PATTERN_TOO_LARGE = -10003,
REGEX_PCRE2_ERROR_CONFIG = -10004,
REGEX_PCRE2_ERROR_SUBJECT_TOO_LARGE = -10005,
REGEX_PCRE2_ERROR_REPLACEMENT_TOO_LARGE = -10006,
REGEX_PCRE2_ERROR_LIMIT_OUT_OF_RANGE = -10007,
REGEX_PCRE2_ERROR_CAPTURE_LIMIT = -10008,
REGEX_PCRE2_ERROR_OUTPUT_BUFFER = -10009
};
enum regex_pcre2_error_phase {
REGEX_PCRE2_PHASE_NONE = 0,
REGEX_PCRE2_PHASE_BRIDGE = 1,
REGEX_PCRE2_PHASE_COMPILE = 2,
REGEX_PCRE2_PHASE_MATCH = 3,
REGEX_PCRE2_PHASE_SUBSTITUTE = 4
};
/*
* This 20-byte, five-u32 record is retained for deterministic compile-only
* build verification. Runtime callers should use the bounded APIs below.
*/
typedef struct regex_pcre2_compile_result {
int32_t status;
uint32_t error_offset;
uint32_t capture_count;
uint32_t name_count;
uint32_t effective_options;
} regex_pcre2_compile_result;
/* All values are required and validated against the bridge hard maxima. */
typedef struct regex_pcre2_limits {
uint32_t maximum_matches;
uint32_t maximum_capture_rows;
uint32_t match_limit;
uint32_t depth_limit;
uint32_t heap_limit_kib;
} regex_pcre2_limits;
/*
* One record is emitted for group zero and every capture group of each
* retained match. Unset captures use REGEX_PCRE2_UNSET_OFFSET for both bounds.
*/
typedef struct regex_pcre2_match_record {
uint32_t match_number;
uint32_t group_number;
uint32_t start_byte;
uint32_t end_byte;
} regex_pcre2_match_record;
/* Names are UTF-8 slices into the caller-owned name byte buffer. */
typedef struct regex_pcre2_name_record {
uint32_t group_number;
uint32_t name_offset;
uint32_t name_length;
} regex_pcre2_name_record;
/*
* Trace collection has an independent bound from match materialization. The
* byte cap covers the fixed event records plus copied mark bytes.
*/
typedef struct regex_pcre2_trace_limits {
uint32_t maximum_events;
uint32_t maximum_trace_bytes;
uint32_t match_limit;
uint32_t depth_limit;
uint32_t heap_limit_kib;
} regex_pcre2_trace_limits;
/*
* Every field except the mark slice is copied directly from a PCRE2 callout
* block. Positions and lengths are offsets in the 8-bit UTF-8 pattern or
* subject supplied to the engine.
*/
typedef struct regex_pcre2_trace_event {
uint32_t callout_number;
uint32_t pattern_position_byte;
uint32_t next_item_length_byte;
uint32_t subject_position_byte;
uint32_t capture_top;
uint32_t capture_last;
uint32_t mark_offset;
uint32_t mark_length;
} regex_pcre2_trace_event;
/*
* A zero status means that a bounded trace was returned. native_match_status
* retains pcre2_match()'s exact result, including PCRE2_ERROR_CALLOUT when the
* bridge deliberately stopped at a trace cap. last_complete_event is
* UINT32_MAX when no complete event was copied.
*/
typedef struct regex_pcre2_trace_result {
int32_t status;
uint32_t error_phase;
uint32_t error_offset;
int32_t native_match_status;
uint32_t effective_options;
uint32_t event_count;
uint32_t total_event_count;
uint32_t mark_bytes_length;
uint32_t trace_bytes_length;
uint32_t events_truncated;
uint32_t marks_truncated;
uint32_t last_complete_event;
uint32_t matched;
} regex_pcre2_trace_result;
/*
* This fixed 60-byte record contains no native pointers. `status` is zero on
* success. Otherwise `error_phase` distinguishes bridge validation, compile,
* match and replacement errors; `error_offset` is a UTF-8 byte offset when the
* phase provides one.
*/
typedef struct regex_pcre2_run_result {
int32_t status;
uint32_t error_phase;
uint32_t error_offset;
uint32_t capture_count;
uint32_t name_count;
uint32_t match_count;
uint32_t record_count;
uint32_t name_record_count;
uint32_t name_bytes_length;
uint32_t effective_options;
uint32_t results_truncated;
uint32_t names_truncated;
uint32_t output_length;
uint32_t substitution_count;
uint32_t output_truncated;
} regex_pcre2_run_result;
uint32_t regex_pcre2_bridge_abi_version(void);
uint32_t regex_pcre2_config_flags(void);
const uint8_t *regex_pcre2_version(void);
int32_t regex_pcre2_error_message(int32_t error_code,
uint8_t *output,
uint32_t output_capacity);
int32_t regex_pcre2_compile_probe(const uint8_t *pattern,
uint32_t pattern_length,
uint32_t application_flags,
regex_pcre2_compile_result *result);
int32_t regex_pcre2_execute(
const uint8_t *pattern,
uint32_t pattern_length,
const uint8_t *subject,
uint32_t subject_length,
uint32_t application_flags,
const regex_pcre2_limits *limits,
regex_pcre2_match_record *records,
uint32_t record_capacity,
regex_pcre2_name_record *name_records,
uint32_t name_record_capacity,
uint8_t *name_bytes,
uint32_t name_bytes_capacity,
regex_pcre2_run_result *result);
int32_t regex_pcre2_substitute(
const uint8_t *pattern,
uint32_t pattern_length,
const uint8_t *subject,
uint32_t subject_length,
const uint8_t *replacement,
uint32_t replacement_length,
uint32_t application_flags,
const regex_pcre2_limits *limits,
regex_pcre2_match_record *records,
uint32_t record_capacity,
regex_pcre2_name_record *name_records,
uint32_t name_record_capacity,
uint8_t *name_bytes,
uint32_t name_bytes_capacity,
/* `output` must have output_capacity + 1 writable bytes for PCRE2's NUL. */
uint8_t *output,
uint32_t output_capacity,
regex_pcre2_run_result *result);
int32_t regex_pcre2_trace(
const uint8_t *pattern,
uint32_t pattern_length,
const uint8_t *subject,
uint32_t subject_length,
uint32_t application_flags,
const regex_pcre2_trace_limits *limits,
regex_pcre2_trace_event *events,
uint32_t event_capacity,
uint8_t *mark_bytes,
uint32_t mark_bytes_capacity,
regex_pcre2_trace_result *result);
int32_t regex_pcre2_self_test(void);
#endif