Files
util.nix/package/c/hemar/hemar.c

1525 lines
50 KiB
C
Executable File

/*
* hemar.c
* Template parser implementation for Hemar
*/
#include "hemar.h"
#include "postgres.h"
#include <string.h>
#include <ctype.h>
#include "utils/elog.h"
#include "utils/memutils.h"
#include "mb/pg_wchar.h"
#include "fmgr.h"
#include "utils/builtins.h"
#include "lib/stringinfo.h"
#include "utils/jsonb.h"
#include "catalog/pg_type.h"
#include "executor/spi.h"
#include "funcapi.h"
/* Forward declarations */
static void template_node_to_string(TemplateNode *node, StringInfo result, int indent);
static const char *
jbt_type_to_string(JsonbIteratorToken type)
{
switch (type)
{
case WJB_DONE:
return "WJB_DONE";
case WJB_KEY:
return "WJB_KEY";
case WJB_VALUE:
return "WJB_VALUE";
case WJB_ELEM:
return "WJB_ELEM";
case WJB_BEGIN_ARRAY:
return "WJB_BEGIN_ARRAY";
case WJB_END_ARRAY:
return "WJB_END_ARRAY";
case WJB_BEGIN_OBJECT:
return "WJB_BEGIN_OBJECT";
case WJB_END_OBJECT:
return "WJB_END_OBJECT";
default:
return "Unknown";
}
}
static char *
jbv_type_to_string(enum jbvType type)
{
switch (type)
{
case jbvNull:
return "jbvNull";
case jbvString:
return "jbvString";
case jbvNumeric:
return "jbvNumeric";
case jbvBool:
return "jbvBool";
case jbvArray:
return "jbvArray";
case jbvObject:
return "jbvObject";
case jbvBinary:
return "jbvBinary";
case jbvDatetime:
return "jbvDatetime";
default:
return "Unknown";
}
}
/* Helper function to skip whitespace */
static const char *
skip_whitespace(const char *s)
{
while (*s && isspace((unsigned char)*s))
s++;
return s;
}
/* Initialize template value based on type */
static TemplateValue
init_template_value(TemplateNodeType type)
{
TemplateValue value;
switch (type)
{
case TEMPLATE_NODE_TEXT:
value.text.content = NULL;
break;
case TEMPLATE_NODE_INTERPOLATE:
value.interpolate.key = NULL;
break;
case TEMPLATE_NODE_SECTION:
value.section.iterator = NULL;
value.section.collection = NULL;
value.section.body = NULL;
break;
case TEMPLATE_NODE_EXECUTE:
value.execute.code = NULL;
break;
case TEMPLATE_NODE_INCLUDE:
value.include.key = NULL;
break;
default:
elog(ERROR, "Unknown template node type: %d", type);
/* This won't be reached due to elog(ERROR) */
memset(&value, 0, sizeof(value));
break;
}
return value;
}
/* Initialize a new template node */
static TemplateNode *
init_template_node(MemoryContext context, TemplateNodeType type)
{
TemplateNode *node;
node = (TemplateNode *) MemoryContextAlloc(context, sizeof(TemplateNode));
node->next = NULL;
node->type = type;
node->value = (TemplateValue *) MemoryContextAlloc(context, sizeof(TemplateValue));
*node->value = init_template_value(type);
return node;
}
/* Error code to string conversion */
const char *
template_error_to_string(TemplateErrorCode code, TemplateConfig *config)
{
char *message = "";
switch (code)
{
case TEMPLATE_ERROR_NONE:
return "No error";
case TEMPLATE_ERROR_UNKNOWN_TAG:
return "Unknown tag";
case TEMPLATE_ERROR_NESTED_INTERPOLATION:
return "Nested interpolation";
case TEMPLATE_UNEXPECTED_OPEN_BRACES_AFFTER_SECTION_CONTROLE:
message = "Found `";
strcat(message, config->Syntax.Braces.open);
strcat(message, "` in `");
strcat(message, config->Syntax.Section.control);
strcat(message, "` in section block");
return message;
case TEMPLATE_UNEXPECTED_OPEN_BRACES_AFFTER_SECTION_SOURCE:
message = "Found `";
strcat(message, config->Syntax.Braces.open);
strcat(message, "` in `");
strcat(message, config->Syntax.Section.source);
strcat(message, "` in section block");
return message;
case TEMPLATE_ERROR_UNEXPECTED_INTERPOLATION_END:
return "Unexpected interpolation end";
case TEMPLATE_ERROR_NO_SOURSE_IN_SECTION:
message = "Not found `";
strcat(message, config->Syntax.Section.source);
strcat(message, "` keyword in section block");
return message;
case TEMPLATE_ERROR_NO_BEGIN_IN_SECTION:
message = "Not found `";
strcat(message, config->Syntax.Section.begin);
strcat(message, "` keyword in section block");
return message;
case TEMPLATE_ERROR_UNEXPECTED_SECTION_END:
return "Unexpected section end";
case TEMPLATE_ERROR_NESTED_INCLUDE:
return "Nested include";
case TEMPLATE_ERROR_NESTED_EXECUTE:
return "Nested execute";
case TEMPLATE_ERROR_INVALID_CONFIG:
return "Invalid config";
case TEMPLATE_ERROR_OUT_OF_MEMORY:
return "Out of memory";
case TEMPLATE_ERROR_UNEXPECTED_INCLUDE_END:
return "Unexpected include end";
case TEMPLATE_ERROR_UNEXPECTED_EXECUTE_END:
return "Unexpected execute end";
default:
return "Unknown error";
}
}
/* Default template configuration */
TemplateConfig
template_default_config(MemoryContext context)
{
TemplateConfig config;
config.Syntax.Braces.open = "{{";
config.Syntax.Braces.close = "}}";
config.Syntax.Section.control = "for ";
config.Syntax.Section.source = "in ";
config.Syntax.Section.begin = "do ";
config.Syntax.Interpolate.invoke = "";
config.Syntax.Include.invoke = "include ";
config.Syntax.Execute.invoke = "exec ";
config.Syntax.nesting = "->";
return config;
}
/* Validate template configuration */
bool
template_validate_config(const TemplateConfig *config, TemplateErrorCode *error_code)
{
if (!config)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check open brace */
if (!config->Syntax.Braces.open || strlen(config->Syntax.Braces.open) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check close brace */
if (!config->Syntax.Braces.close || strlen(config->Syntax.Braces.close) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check section control */
if (!config->Syntax.Section.control || strlen(config->Syntax.Section.control) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check section source */
if (!config->Syntax.Section.source || strlen(config->Syntax.Section.source) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check section begin */
if (!config->Syntax.Section.begin || strlen(config->Syntax.Section.begin) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check interpolate invoke */
if (!config->Syntax.Interpolate.invoke || strlen(config->Syntax.Interpolate.invoke) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check include invoke */
if (!config->Syntax.Include.invoke || strlen(config->Syntax.Include.invoke) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check execute invoke */
if (!config->Syntax.Execute.invoke || strlen(config->Syntax.Execute.invoke) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
/* Check nesting */
if (!config->Syntax.nesting || strlen(config->Syntax.nesting) > TEMPLATE_MAX_PREFIX_LEN)
{
if (error_code)
*error_code = TEMPLATE_ERROR_INVALID_CONFIG;
return false;
}
if (error_code)
*error_code = TEMPLATE_ERROR_NONE;
return true;
}
/* Parse interpolation tag */
static TemplateNode *
template_parse_interpolation(MemoryContext context, const char **s_ptr,
const TemplateConfig *config, TemplateErrorCode *error_code)
{
const char **s = s_ptr;
const char *key_start;
size_t key_len;
TemplateNode *node;
node = init_template_node(context, TEMPLATE_NODE_INTERPOLATE);
/* Skip to the content */
*s += strlen(config->Syntax.Braces.open);
*s = skip_whitespace(*s);
*s += strlen(config->Syntax.Interpolate.invoke);
*s = skip_whitespace(*s);
key_start = *s;
while (**s != '\0')
{
if (isspace((unsigned char)**s) ||
strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) == 0)
break;
if (strncmp(*s, config->Syntax.Braces.open, strlen(config->Syntax.Braces.open)) == 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_NESTED_INTERPOLATION;
template_free_node(node);
return NULL;
}
(*s)++;
}
key_len = *s - key_start;
node->value->interpolate.key = MemoryContextStrdup(context, pnstrdup(key_start, key_len));
elog(DEBUG1, "TPI: Parsing: %s", node->value->interpolate.key);
*s = skip_whitespace(*s);
/* Check for closing brace */
if (strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) != 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_UNEXPECTED_INTERPOLATION_END;
template_free_node(node);
return NULL;
}
*s_ptr = *s + strlen(config->Syntax.Braces.close);
return node;
}
/* Parse section tag */
static TemplateNode *
template_parse_section(MemoryContext context, const char **s_ptr,
const TemplateConfig *config, TemplateErrorCode *error_code)
{
const char **s = s_ptr;
const char *iterator_start, *collection_start;
size_t iterator_len, collection_len;
TemplateNode *node, *body_node;
node = init_template_node(context, TEMPLATE_NODE_SECTION);
/* Skip to the content */
*s += strlen(config->Syntax.Braces.open);
*s = skip_whitespace(*s);
*s += strlen(config->Syntax.Section.control);
/* Find the iterator name */
*s = skip_whitespace(*s);
iterator_start = *s;
while (**s != '\0')
{
if (isspace((unsigned char)**s) ||
strncmp(*s, config->Syntax.Section.source, strlen(config->Syntax.Section.source)) == 0)
break;
if (strncmp(*s, config->Syntax.Braces.open, strlen(config->Syntax.Braces.open)) == 0)
{
if (error_code)
*error_code = TEMPLATE_UNEXPECTED_OPEN_BRACES_AFFTER_SECTION_CONTROLE;
template_free_node(node);
return NULL;
}
if (strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) == 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_NO_SOURSE_IN_SECTION;
template_free_node(node);
return NULL;
}
(*s)++;
}
iterator_len = *s - iterator_start;
node->value->section.iterator = MemoryContextStrdup(context, pnstrdup(iterator_start, iterator_len));
elog(DEBUG1, "TPS: Parsed section iterator: %s", node->value->section.iterator);
/* Find the collection name */
*s = skip_whitespace(*s);
if (strncmp(*s, config->Syntax.Section.source, strlen(config->Syntax.Section.source)) != 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_NO_SOURSE_IN_SECTION;
template_free_node(node);
return NULL;
}
*s += strlen(config->Syntax.Section.source);
*s = skip_whitespace(*s);
collection_start = *s;
while (**s != '\0')
{
if (isspace((unsigned char)**s) ||
strncmp(*s, config->Syntax.Section.begin, strlen(config->Syntax.Section.begin)) == 0)
break;
if (strncmp(*s, config->Syntax.Braces.open, strlen(config->Syntax.Braces.open)) == 0)
{
if (error_code)
*error_code = TEMPLATE_UNEXPECTED_OPEN_BRACES_AFFTER_SECTION_SOURCE;
template_free_node(node);
return NULL;
}
if (strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) == 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_NO_BEGIN_IN_SECTION;
template_free_node(node);
return NULL;
}
(*s)++;
}
collection_len = *s - collection_start;
node->value->section.collection = MemoryContextStrdup(context, pnstrdup(collection_start, collection_len));
elog(DEBUG1, "TPS: Parsed section collection: %s", node->value->section.collection);
/* Check for 'do' keyword */
*s = skip_whitespace(*s);
// TODO: why check begin second time, first in while
if (strncmp(*s, config->Syntax.Section.begin, strlen(config->Syntax.Section.begin)) != 0)
{
if (error_code)
*error_code = TEMPLATE_UNEXPECTED_OPEN_BRACES_AFFTER_SECTION_SOURCE;
template_free_node(node);
return NULL;
}
*s += strlen(config->Syntax.Section.begin);
*s = skip_whitespace(*s);
/* Check if there's a closing brace right after 'do' */
if (strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) == 0)
{
/* Empty section body */
elog(DEBUG1, "TPS: Parsed empty section body");
*s_ptr = *s + strlen(config->Syntax.Braces.close);
node->value->section.body = NULL;
return node;
}
/* Parse the body as a normal template */
const char *body_start = *s;
const char *original_s = *s;
int inner_braces_opened_count = 0;
/* Find the end of the section */
while (**s) {
// s = {% a %} %}
elog(DEBUG2, "TPS: Step, braces opened: %d, s: %s", inner_braces_opened_count, *s);
if (strncmp(*s, config->Syntax.Braces.open, strlen(config->Syntax.Braces.open)) == 0) {
elog(DEBUG2, "TPS: inner_braces_opened_count++");
inner_braces_opened_count++;
}
if (strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) == 0) {
if (inner_braces_opened_count > 0) {
elog(DEBUG2, "TPS: inner_braces_opened_count--");
inner_braces_opened_count--;
}
else {
elog(DEBUG2, "TPS: exit");
break;
}
}
(*s)++;
}
if (!**s)
{
/* Unexpected end of string before closing brace */
if (error_code)
*error_code = TEMPLATE_ERROR_UNEXPECTED_SECTION_END;
template_free_node(node);
return NULL;
}
/* Extract the body content */
size_t body_len = *s - body_start;
char *body_content = pnstrdup(body_start, body_len);
elog(DEBUG1, "TPS: Section body content: %s", body_content);
/* Parse the body content as a template */
const char *body_ptr = body_content;
body_node = template_parse(context, &body_ptr, config, false, error_code);
if (!body_node)
{
elog(WARNING, "TPS: Failed to parse section body: %s", body_content);
pfree(body_content);
template_free_node(node);
return NULL;
}
pfree(body_content);
node->value->section.body = body_node;
/* Set the pointer to after the closing brace */
*s_ptr = *s + strlen(config->Syntax.Braces.close);
return node;
}
/* Parse include tag */
static TemplateNode *
template_parse_include(MemoryContext context, const char **s_ptr,
const TemplateConfig *config, TemplateErrorCode *error_code)
{
const char **s = s_ptr;
const char *include_start;
size_t include_len;
TemplateNode *node;
node = init_template_node(context, TEMPLATE_NODE_INCLUDE);
/* Skip to the content */
*s += strlen(config->Syntax.Braces.open);
*s = skip_whitespace(*s);
*s += strlen(config->Syntax.Include.invoke);
*s = skip_whitespace(*s);
include_start = *s;
while (**s != '\0')
{
if (isspace((unsigned char)**s) ||
strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) == 0)
break;
if (strncmp(*s, config->Syntax.Braces.open, strlen(config->Syntax.Braces.open)) == 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_NESTED_INCLUDE;
template_free_node(node);
return NULL;
}
(*s)++;
}
include_len = *s - include_start;
node->value->include.key = MemoryContextStrdup(context, pnstrdup(include_start, include_len));
*s = skip_whitespace(*s);
/* Check for closing brace */
if (strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) != 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_UNEXPECTED_INCLUDE_END;
template_free_node(node);
return NULL;
}
*s_ptr = *s + strlen(config->Syntax.Braces.close);
return node;
}
/* Parse execute tag */
static TemplateNode *
template_parse_execute(MemoryContext context, const char **s_ptr,
const TemplateConfig *config, TemplateErrorCode *error_code)
{
const char **s = s_ptr;
const char *code_start;
size_t code_len;
TemplateNode *node;
node = init_template_node(context, TEMPLATE_NODE_EXECUTE);
/* Skip to the content */
*s += strlen(config->Syntax.Braces.open);
*s = skip_whitespace(*s);
*s += strlen(config->Syntax.Execute.invoke);
*s = skip_whitespace(*s);
code_start = *s;
while (**s != '\0')
{
if (strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) == 0)
break;
if (strncmp(*s, config->Syntax.Braces.open, strlen(config->Syntax.Braces.open)) == 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_NESTED_EXECUTE;
template_free_node(node);
return NULL;
}
(*s)++;
}
code_len = *s - code_start;
node->value->execute.code = MemoryContextStrdup(context, pnstrdup(code_start, code_len));
/* Check for closing brace */
if (strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) != 0)
{
if (error_code)
*error_code = TEMPLATE_ERROR_UNEXPECTED_EXECUTE_END;
template_free_node(node);
return NULL;
}
*s_ptr = *s + strlen(config->Syntax.Braces.close);
return node;
}
/* Main template parser function */
TemplateNode *
template_parse(MemoryContext context, const char **s, const TemplateConfig *config,
bool inner_parse, TemplateErrorCode *error_code)
{
const char *start;
TemplateNode *root, *current, *tag_node;
bool current_node_filled = false;
const char *tag_prefix;
size_t text_len;
if (!template_validate_config(config, error_code))
{
return NULL;
}
/* Check for empty template */
if (!s || !*s || !**s)
{
TemplateNode *empty_node = init_template_node(context, TEMPLATE_NODE_TEXT);
empty_node->value->text.content = MemoryContextStrdup(context, "");
if (error_code)
*error_code = TEMPLATE_ERROR_NONE;
return empty_node;
}
start = *s;
root = init_template_node(context, TEMPLATE_NODE_TEXT);
current = root;
while (*s && **s != '\0')
{
/* Check for closing brace in inner parse */
if (inner_parse && strncmp(*s, config->Syntax.Braces.close, strlen(config->Syntax.Braces.close)) == 0)
{
break;
}
if (strncmp(*s, config->Syntax.Braces.open, strlen(config->Syntax.Braces.open)) == 0)
{
/* Handle text before tag */
if (start != *s)
{
if (current_node_filled)
{
TemplateNode *new_node = init_template_node(context, TEMPLATE_NODE_TEXT);
current->next = new_node;
current = new_node;
}
else
{
current->type = TEMPLATE_NODE_TEXT;
*current->value = init_template_value(TEMPLATE_NODE_TEXT);
}
text_len = *s - start;
current->value->text.content = MemoryContextStrdup(context, pnstrdup(start, text_len));
current_node_filled = true;
}
/* Parse the tag */
tag_node = NULL;
tag_prefix = *s + strlen(config->Syntax.Braces.open);
tag_prefix = skip_whitespace(tag_prefix);
/* Determine tag type by prefix */
if (strncmp(tag_prefix, config->Syntax.Section.control, strlen(config->Syntax.Section.control)) == 0)
{
tag_node = template_parse_section(context, s, config, error_code);
}
else if (strncmp(tag_prefix, config->Syntax.Include.invoke, strlen(config->Syntax.Include.invoke)) == 0)
{
tag_node = template_parse_include(context, s, config, error_code);
}
else if (strncmp(tag_prefix, config->Syntax.Execute.invoke, strlen(config->Syntax.Execute.invoke)) == 0)
{
tag_node = template_parse_execute(context, s, config, error_code);
}
else if (strncmp(tag_prefix, config->Syntax.Interpolate.invoke, strlen(config->Syntax.Interpolate.invoke)) == 0)
{
tag_node = template_parse_interpolation(context, s, config, error_code);
}
else
{
if (error_code)
*error_code = TEMPLATE_ERROR_UNKNOWN_TAG;
template_free_node(root);
return NULL;
}
if (!tag_node)
{
template_free_node(root);
return NULL;
}
if (current_node_filled)
{
current->next = tag_node;
current = tag_node;
}
else
{
*current = *tag_node;
pfree(tag_node);
}
current_node_filled = true;
start = *s;
}
else
{
(*s)++;
}
}
/* Handle remaining text */
if (start != *s)
{
if (current_node_filled)
{
TemplateNode *new_node = init_template_node(context, TEMPLATE_NODE_TEXT);
current->next = new_node;
current = new_node;
}
else
{
current->type = TEMPLATE_NODE_TEXT;
*current->value = init_template_value(TEMPLATE_NODE_TEXT);
}
text_len = *s - start;
current->value->text.content = MemoryContextStrdup(context, pnstrdup(start, text_len));
current_node_filled = true;
}
/* If no nodes were created, ensure we have at least an empty text node */
if (!current_node_filled)
{
current->type = TEMPLATE_NODE_TEXT;
*current->value = init_template_value(TEMPLATE_NODE_TEXT);
current->value->text.content = MemoryContextStrdup(context, "");
}
if (error_code)
*error_code = TEMPLATE_ERROR_NONE;
return root;
}
/* Free a template node and all its children */
void
template_free_node(TemplateNode *node)
{
TemplateNode *current, *next;
if (!node)
return;
current = node;
while (current)
{
next = current->next;
if (current->value)
{
switch (current->type)
{
case TEMPLATE_NODE_TEXT:
if (current->value->text.content)
pfree(current->value->text.content);
break;
case TEMPLATE_NODE_INTERPOLATE:
if (current->value->interpolate.key)
pfree(current->value->interpolate.key);
break;
case TEMPLATE_NODE_SECTION:
if (current->value->section.iterator)
pfree(current->value->section.iterator);
if (current->value->section.collection)
pfree(current->value->section.collection);
if (current->value->section.body)
template_free_node(current->value->section.body);
break;
case TEMPLATE_NODE_EXECUTE:
if (current->value->execute.code)
pfree(current->value->execute.code);
break;
case TEMPLATE_NODE_INCLUDE:
if (current->value->include.key)
pfree(current->value->include.key);
break;
default:
/* Should not happen */
break;
}
pfree(current->value);
}
pfree(current);
current = next;
}
}
/* PostgreSQL function wrappers */
PG_MODULE_MAGIC;
PG_FUNCTION_INFO_V1(pg_template_parse);
Datum
pg_template_parse(PG_FUNCTION_ARGS)
{
text *template_text = PG_GETARG_TEXT_PP(0);
char *template_str = text_to_cstring(template_text);
const char *template_ptr = template_str;
MemoryContext old_context;
MemoryContext parse_context;
TemplateConfig config;
TemplateNode *root = NULL;
TemplateErrorCode error_code = TEMPLATE_ERROR_NONE;
text *result_text = NULL;
StringInfoData result;
/* Create a memory context for parsing */
parse_context = AllocSetContextCreate(CurrentMemoryContext,
"Template Parse Context",
ALLOCSET_DEFAULT_SIZES);
/* Switch to the new context for parsing */
old_context = MemoryContextSwitchTo(parse_context);
/* Initialize default config */
config = template_default_config(parse_context);
PG_TRY();
{
/* Parse the template */
root = template_parse(parse_context, &template_ptr, &config, false, &error_code);
/* Check for parsing errors */
if (error_code != TEMPLATE_ERROR_NONE || !root)
{
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("Template parsing error: %s", template_error_to_string(error_code, &config))));
}
/* Convert the parsed template to a string representation for debugging */
initStringInfo(&result);
appendStringInfo(&result, "Template parsed successfully. Structure:\n");
template_node_to_string(root, &result, 0);
/* Switch back to the original memory context */
MemoryContextSwitchTo(old_context);
/* Return the result */
result_text = cstring_to_text(result.data);
pfree(result.data);
}
PG_CATCH();
{
/* Switch back to the original memory context for error handling */
MemoryContextSwitchTo(old_context);
/* Clean up */
if (template_str)
pfree(template_str);
/* Delete the parse context */
MemoryContextDelete(parse_context);
PG_RE_THROW();
}
PG_END_TRY();
/* Clean up */
MemoryContextDelete(parse_context);
pfree(template_str);
PG_RETURN_TEXT_P(result_text);
}
/* Helper function to convert template node to string for debugging */
static void
template_node_to_string(TemplateNode *node, StringInfo result, int indent)
{
TemplateNode *current = node;
int i;
while (current)
{
/* Add indentation */
for (i = 0; i < indent; i++)
appendStringInfoChar(result, ' ');
/* Add node type */
switch (current->type)
{
case TEMPLATE_NODE_TEXT:
appendStringInfo(result, "TEXT: \"%s\"\n",
current->value->text.content ? current->value->text.content : "");
break;
case TEMPLATE_NODE_INTERPOLATE:
appendStringInfo(result, "INTERPOLATE: \"%s\"\n",
current->value->interpolate.key ? current->value->interpolate.key : "");
break;
case TEMPLATE_NODE_SECTION:
appendStringInfo(result, "SECTION: iterator=\"%s\", collection=\"%s\"\n",
current->value->section.iterator ? current->value->section.iterator : "",
current->value->section.collection ? current->value->section.collection : "");
if (current->value->section.body)
{
template_node_to_string(current->value->section.body, result, indent + 2);
}
break;
case TEMPLATE_NODE_EXECUTE:
appendStringInfo(result, "EXECUTE: \"%s\"\n",
current->value->execute.code ? current->value->execute.code : "");
break;
case TEMPLATE_NODE_INCLUDE:
appendStringInfo(result, "INCLUDE: \"%s\"\n",
current->value->include.key ? current->value->include.key : "");
break;
default:
appendStringInfo(result, "UNKNOWN NODE TYPE: %d\n", current->type);
break;
}
current = current->next;
}
}
/* Function declarations */
PG_FUNCTION_INFO_V1(pg_jsonb_get_by_path);
/*
* Parse a path string into segments for JSONB traversal
* Path format: field.nested_field[0].array_field[1][2]
* Returns an array of segments and the count of segments
*/
static char **
parse_path_string(const char *path_str, int *num_segments)
{
char **segments;
int max_segments = 32; /* Initial capacity */
int segment_count = 0;
StringInfoData current_segment;
const char *p = path_str;
bool in_brackets = false;
/* Allocate memory for segments */
segments = (char **)palloc(sizeof(char *) * max_segments);
if (segments == NULL)
ereport(ERROR,
(errcode(ERRCODE_OUT_OF_MEMORY),
errmsg("out of memory")));
initStringInfo(&current_segment);
/* Parse the path string */
while (*p)
{
if (*p == '.' && !in_brackets)
{
/* End of segment */
if (current_segment.len > 0)
{
segments[segment_count++] = pstrdup(current_segment.data);
resetStringInfo(&current_segment);
/* Resize if needed */
if (segment_count >= max_segments)
{
max_segments *= 2;
segments = (char **)repalloc(segments, sizeof(char *) * max_segments);
if (segments == NULL)
ereport(ERROR,
(errcode(ERRCODE_OUT_OF_MEMORY),
errmsg("out of memory")));
}
}
}
else if (*p == '[')
{
/* Start of array index */
if (in_brackets)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid path syntax: nested brackets")));
/* Save the current field name if any */
if (current_segment.len > 0)
{
segments[segment_count++] = pstrdup(current_segment.data);
resetStringInfo(&current_segment);
/* Resize if needed */
if (segment_count >= max_segments)
{
max_segments *= 2;
segments = (char **)repalloc(segments, sizeof(char *) * max_segments);
if (segments == NULL)
ereport(ERROR,
(errcode(ERRCODE_OUT_OF_MEMORY),
errmsg("out of memory")));
}
}
/* Add a special prefix to indicate this is an array index */
appendStringInfoChar(&current_segment, '[');
in_brackets = true;
}
else if (*p == ']')
{
/* End of array index */
if (!in_brackets)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid path syntax: unmatched closing bracket")));
/* Save the array index */
if (current_segment.len > 0)
{
/* We don't need to append the closing bracket, just save the segment */
segments[segment_count++] = pstrdup(current_segment.data);
resetStringInfo(&current_segment);
/* Resize if needed */
if (segment_count >= max_segments)
{
max_segments *= 2;
segments = (char **)repalloc(segments, sizeof(char *) * max_segments);
if (segments == NULL)
ereport(ERROR,
(errcode(ERRCODE_OUT_OF_MEMORY),
errmsg("out of memory")));
}
}
else
{
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid path syntax: empty array index")));
}
in_brackets = false;
}
else
{
/* Regular character */
appendStringInfoChar(&current_segment, *p);
}
p++;
}
/* Handle the last segment if any */
if (current_segment.len > 0)
{
if (in_brackets)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid path syntax: unclosed bracket")));
segments[segment_count++] = pstrdup(current_segment.data);
}
pfree(current_segment.data);
*num_segments = segment_count;
return segments;
}
/*
* Get JSONB value by path
* Input: JSONB document and path string
* Output: JSONB value at the specified path or NULL if path is invalid
*/
Datum
pg_jsonb_get_by_path(PG_FUNCTION_ARGS)
{
Jsonb *jb = PG_GETARG_JSONB_P(0);
text *path_text = PG_GETARG_TEXT_PP(1);
char *path_str = text_to_cstring(path_text);
JsonbValue *result = NULL;
JsonbValue tmp_val;
JsonbIterator *it;
JsonbIteratorToken token;
JsonbValue v;
char **segments;
int num_segments;
int current_segment = 0;
int array_index;
int current_index;
bool found;
int i;
/* Empty path returns NULL */
if (path_str == NULL || *path_str == '\0')
{
elog(WARNING, "path is empty, returning NULL");
pfree(path_str);
PG_RETURN_NULL();
}
/* Parse the path string */
segments = parse_path_string(path_str, &num_segments);
if (num_segments == 0)
{
/* Free allocated memory before returning */
elog(WARNING, "no segments in path, returning NULL");
pfree(path_str);
pfree(segments);
PG_RETURN_NULL();
}
/* Start iterating through the JSONB document */
it = JsonbIteratorInit(&jb->root);
token = JsonbIteratorNext(&it, &v, true);
if (token == WJB_BEGIN_ARRAY)
{
/* If the root is an array, the first segment must be an array index */
if (current_segment < num_segments && segments[current_segment][0] != '[')
{
/* First segment is not an array index, but root is an array */
/* Free allocated memory before returning */
elog(WARNING, "root is an array, but first segment '%s' is not an array index", segments[current_segment]);
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
}
else if (token == WJB_BEGIN_OBJECT)
{
/* If the root is an object, the first segment must be a field name */
if (current_segment < num_segments && segments[current_segment][0] == '[')
{
/* First segment is an array index, but root is an object */
/* Free allocated memory before returning */
elog(WARNING, "root is an object, but first segment '%s' is an array index", segments[current_segment]);
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
}
else
{
/* Root is a scalar value, can't navigate further */
/* Free allocated memory before returning */
elog(WARNING, "root is a scalar value, can't navigate further");
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
/* Process each path segment */
while (current_segment < num_segments)
{
char *segment = segments[current_segment];
if (segment[0] == '[')
{
/* Array index segment */
if (token != WJB_BEGIN_ARRAY)
{
/* Not an array, can't use array index */
/* Free allocated memory before returning */
elog(WARNING, "segment '%s' is not an array, can't use array index", segment);
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
/* Extract the array index */
array_index = atoi(segment + 1);
if (array_index < 0)
{
/* Free allocated memory before returning */
elog(WARNING, "invalid array index: %d", array_index);
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
/* Navigate to the specified array element */
current_index = 0;
found = false;
/* Move to the first element */
token = JsonbIteratorNext(&it, &v, true);
/* Iterate through array elements */
while (token != WJB_END_ARRAY)
{
if (current_index == array_index)
{
found = true;
/* If this is the last segment, return the value */
if (current_segment == num_segments - 1)
{
/* Create a standalone JsonbValue for the result */
result = palloc(sizeof(JsonbValue));
*result = v;
/* Convert to a Jsonb container */
if (result->type == jbvBinary)
{
tmp_val.type = jbvBinary;
tmp_val.val.binary.data = result->val.binary.data;
tmp_val.val.binary.len = result->val.binary.len;
}
else
{
tmp_val = *result;
}
/* Free allocated memory before returning */
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
/* Return the result as a new Jsonb */
PG_RETURN_JSONB_P(JsonbValueToJsonb(&tmp_val));
}
/* Not the last segment, continue traversing */
if (v.type == jbvBinary)
{
/* Nested container, need to iterate into it */
JsonbIterator *nested_it = JsonbIteratorInit((JsonbContainer *) v.val.binary.data);
it = nested_it;
/* Get the container type (array or object) */
token = JsonbIteratorNext(&it, &v, true);
/* Move to the next segment */
current_segment++;
/* Continue processing from the next segment */
break;
}
else
{
/* Scalar value, can't navigate further */
/* Free allocated memory before returning */
elog(WARNING, "scalar value at segment '%s', can't navigate further", segment);
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
}
/* Skip this element */
if (v.type == jbvBinary)
{
/* Skip over the entire container */
JsonbIterator *nested_it = JsonbIteratorInit((JsonbContainer *) v.val.binary.data);
JsonbIteratorToken nested_token;
JsonbValue nested_v;
/* Skip until we reach the end of the container */
do
{
nested_token = JsonbIteratorNext(&nested_it, &nested_v, false);
} while (nested_token != WJB_DONE);
}
current_index++;
token = JsonbIteratorNext(&it, &v, true);
}
if (!found)
{
/* Array index out of bounds */
/* Free allocated memory before returning */
elog(WARNING, "array index %d out of bounds", array_index);
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
/* If we found the element and broke out of the loop to process the next segment,
we need to continue the outer loop */
if (current_segment < num_segments)
continue;
}
else
{
/* Field name segment */
if (token != WJB_BEGIN_OBJECT)
{
/* Not an object, can't use field name */
/* Free allocated memory before returning */
elog(WARNING, "segment '%s' is not an object, can't use field name", segment);
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
/* Navigate to the specified field */
found = false;
token = JsonbIteratorNext(&it, &v, true);
while (token != WJB_END_OBJECT)
{
/* We should be at a key */
if (token != WJB_KEY)
{
/* Free allocated memory before returning */
elog(WARNING, "unexpected token: expected WJB_KEY");
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
ereport(ERROR,
(errcode(ERRCODE_INTERNAL_ERROR),
errmsg("unexpected JSONB iterator token")));
}
/* Check if this is the field we're looking for */
if (v.type == jbvString &&
strlen(segment) == v.val.string.len &&
strncmp(segment, v.val.string.val, v.val.string.len) == 0)
{
found = true;
/* Get the value */
token = JsonbIteratorNext(&it, &v, true);
/* If this is the last segment, return the value */
if (current_segment == num_segments - 1)
{
/* Create a standalone JsonbValue for the result */
result = palloc(sizeof(JsonbValue));
*result = v;
/* Convert to a Jsonb container */
if (result->type == jbvBinary)
{
tmp_val.type = jbvBinary;
tmp_val.val.binary.data = result->val.binary.data;
tmp_val.val.binary.len = result->val.binary.len;
}
else
{
tmp_val = *result;
}
/* Free allocated memory before returning */
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
/* Return the result as a new Jsonb */
PG_RETURN_JSONB_P(JsonbValueToJsonb(&tmp_val));
}
/* Not the last segment, continue traversing */
if (v.type == jbvBinary)
{
/* Nested container, need to iterate into it */
JsonbIterator *nested_it = JsonbIteratorInit((JsonbContainer *) v.val.binary.data);
it = nested_it;
/* Get the container type (array or object) */
token = JsonbIteratorNext(&it, &v, true);
/* Move to the next segment */
current_segment++;
/* Continue processing from the next segment */
break;
}
else
{
/* Scalar value, can't navigate further */
/* Free allocated memory before returning */
elog(WARNING, "scalar value at segment '%s', can't navigate further", segment);
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
}
else
{
/* Skip this field's value */
token = JsonbIteratorNext(&it, &v, true);
if (v.type == jbvBinary)
{
/* Skip over the entire container */
JsonbIterator *nested_it = JsonbIteratorInit((JsonbContainer *) v.val.binary.data);
JsonbIteratorToken nested_token;
JsonbValue nested_v;
/* Skip until we reach the end of the container */
do
{
nested_token = JsonbIteratorNext(&nested_it, &nested_v, false);
} while (nested_token != WJB_DONE);
}
}
token = JsonbIteratorNext(&it, &v, true);
}
if (!found)
{
/* Field not found */
/* Free allocated memory before returning */
elog(WARNING, "field not found, returning NULL");
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}
/* If we found the field and broke out of the loop to process the next segment,
we need to continue the outer loop */
if (current_segment < num_segments)
continue;
}
current_segment++;
}
/* We should have returned by now if we found the value */
/* Free allocated memory before returning */
elog(WARNING, "unexpected end of function, returning NULL");
for (i = 0; i < num_segments; i++)
pfree(segments[i]);
pfree(segments);
pfree(path_str);
PG_RETURN_NULL();
}