Word Unperfect - XT-Emporium

File editor

#ifndef WP_CORPUS_VALIDATOR_H
#define WP_CORPUS_VALIDATOR_H

#include <stdbool.h>
#include <stddef.h>
#include <stdint.h>
#include <stdio.h>
#include "wp_resource_manager.h"
#include "wp_variable_codes.h"

typedef struct WpCorpusValidationOptions {
    bool document_streams_only;  /* Validate only WP file_type 0x0A streams. */
    bool print_per_file;         /* Emit one status line per validated file. */
} WpCorpusValidationOptions;

typedef struct WpCorpusValidationStats {
    size_t paths_seen;
    size_t wp_files_seen;
    size_t document_files_seen;
    size_t skipped_wp_files;
    size_t load_failures;
    size_t files_passed;
    size_t files_failed;
    size_t body_bytes;
    size_t records_seen;
    size_t top_level_records_seen;
    size_t top_level_bytes_consumed;
    size_t char_records;
    size_t single_byte_codes;
    size_t fixed_length_codes;
    size_t variable_length_codes;
    size_t ascii_control_records;
    size_t printable_text_records;
    size_t single_byte_format_records;
    size_t packet_records;
    size_t line_break_records;
    size_t page_break_records;
    size_t hard_returns;
    size_t soft_returns;
    size_t hard_pages;
    size_t soft_pages;
    size_t tabs;
    size_t indents;
    size_t dormant_returns;
    size_t dormant_pages;
    size_t normalized_spaces;
    size_t normalized_hyphens;
    size_t whitespace_gate_records;
    size_t render_dirty_gate_records;
    size_t dirty_exempt_records;
    size_t dirty_neutral_records;
    size_t extension_scan_stop_records;
    size_t tsm_highlight_records;
    size_t hyphenation_suppression_records;
    size_t hidden_function_gate_records;
    size_t unknown_single_byte_records;
    size_t display_columns;
    size_t fixed_known_packets;
    size_t fixed_unknown_packets;
    size_t fixed_mirrored_packets;
    size_t fixed_zero_payload_packets;
    size_t extended_character_packets;
    size_t attribute_begin_packets;
    size_t attribute_end_packets;
    size_t d0_initial_format_tables;
    size_t d0_format_table_words;
    size_t d0_format_table_groups;
    size_t d0_format_table_sentinels;
    size_t d0_format_table_nonmonotonic_steps;
    uint16_t d0_format_table_first_step;
    uint16_t d0_format_table_max_value;
    size_t d4_layout_state_records;
    size_t d4_layout_state_payload_bytes;
    size_t d4_pending_span_records;
    size_t d4_line_window_records;
    size_t d4_line_window_extension_bytes;
    size_t d4_line_metric_records;
    size_t d4_line_metric_extension_bytes;
    size_t d4_extension_fragment_records;
    size_t d4_extension_fragment_bytes;
    size_t d4_line_build_checkpoint_records;
    size_t d4_word_pair_checkpoint_records;
    size_t d4_position_marker_records;
    size_t d4_control_word_records;
    size_t d4_compact_metric_records;
    size_t d4_layout_anchor_records;
    size_t d4_unknown_records;
    size_t d4_unknown_payload_bytes;
    size_t d4_trailing_unparsed_bytes;
    size_t d4_extension_block_records;
    size_t d4_extension_block_bytes;
    size_t variable_structural_packets;
    size_t variable_generic_packets;
    size_t d1_definition_packets;
    size_t d2_outline_packets;
    size_t d3_generated_text_packets;
    size_t repeat_group_packets;
    size_t delayed_text_packets;
    size_t box_object_packets;
    size_t table_layout_packets;
    size_t system_command_packets;
    size_t payload_bytes;

    size_t incomplete_records;
    size_t mismatched_trailers;
    size_t unknown_fixed_packets;
    size_t code_histogram[256];
    size_t variable_action_histogram[WP_VARIABLE_ACTION_COUNT];
    size_t nested_stream_hints;
    size_t nested_streams_analyzed;
    size_t nested_stream_records;
    size_t nested_stream_bytes;
    size_t nested_stream_max_depth;
    size_t nested_stream_parse_gaps;
    size_t nested_stream_recursion_limits;
    size_t variable_repeat_dispatch_packets;
    size_t variable_repeat_total;
    size_t variable_extension_scan_packets;
    size_t variable_refcount_increment_packets;
    size_t variable_refcount_decrement_packets;
    size_t scanner_bypass_packets;
} WpCorpusValidationStats;

typedef struct WpCorpusRoundTripStats {
    size_t paths_seen;
    size_t wp_files_seen;
    size_t files_passed;
    size_t files_failed;
    size_t load_failures;
    size_t write_failures;
    size_t compare_failures;
    size_t bytes_compared;
    size_t preservation_validated;
    size_t validation_failures;
    size_t prefix_bytes;
    size_t body_bytes;
    size_t file_type_histogram[256];
} WpCorpusRoundTripStats;

typedef struct WpCorpusInventoryStats {
    size_t paths_seen;
    size_t wp_files_seen;
    size_t files_passed;
    size_t files_failed;
    size_t load_failures;
    size_t bytes_seen;
    size_t prefix_bytes;
    size_t body_bytes;
    size_t prefix_only_files;
    size_t files_with_body;
    size_t macro_files;
    size_t macro_words;
    size_t macro_literal_words;
    size_t macro_control_words;
    size_t macro_command_words;
    size_t macro_extended_words;
    size_t macro_zero_words;
    size_t macro_odd_body_files;
    size_t macro_titled_files;
    size_t macro_title_bytes;
    size_t macro_preview_chars;
    size_t macro_unique_command_opcodes;
    size_t macro_command_histogram[WP_RES_MACRO_COMMAND_OPCODE_COUNT];
    size_t keyboard_files;
    size_t keyboard_slots;
    size_t keyboard_empty_slots;
    size_t keyboard_bound_slots;
    size_t keyboard_literal_refs;
    size_t keyboard_macro_refs;
    size_t keyboard_command_refs;
    size_t keyboard_macro_command_refs;
    size_t keyboard_other_refs;
    size_t keyboard_trailing_bytes;
    size_t keyboard_unique_literal_refs;
    size_t keyboard_unique_macro_refs;
    size_t keyboard_unique_command_refs;
    size_t keyboard_unique_macro_command_refs;
    size_t keyboard_unique_entry_classes;
    size_t keyboard_descriptor_records;
    size_t keyboard_descriptor_bytes;
    size_t keyboard_descriptor_files;
    size_t keyboard_section_entries;
    size_t keyboard_valid_sections;
    size_t keyboard_invalid_sections;
    size_t keyboard_section_payload_bytes;
    size_t keyboard_binding_section_files;
    size_t keyboard_descriptor_section_files;
    uint8_t keyboard_literal_ref_bitmap[WP_RES_BYTE_BITMAP_BYTES];
    uint8_t keyboard_macro_ref_bitmap[WP_RES_BYTE_BITMAP_BYTES];
    uint8_t keyboard_command_ref_bitmap[WP_RES_BYTE_BITMAP_BYTES];
    uint8_t keyboard_macro_command_ref_bitmap[WP_RES_BYTE_BITMAP_BYTES];
    uint8_t keyboard_entry_class_bitmap[WP_RES_BYTE_BITMAP_BYTES];
    size_t printer_files;
    size_t printer_named_files;
    size_t printer_body_words;
    size_t printer_zero_words;
    size_t printer_ffff_words;
    size_t printer_offset_like_words;
    size_t printer_odd_body_files;
    size_t resource_generic_files;
    size_t resource_generic_prefix_bytes;
    size_t resource_generic_body_bytes;
    size_t resource_generic_body_words;
    size_t resource_generic_zero_words;
    size_t resource_generic_ffff_words;
    size_t resource_generic_offset_like_words;
    size_t resource_generic_printable_runs;
    size_t resource_generic_printable_bytes;
    size_t resource_generic_length_strings;
    size_t resource_generic_length_string_bytes;
    size_t resource_generic_string_files;
    size_t resource_generic_odd_body_files;
    size_t file_type_histogram[256];
    size_t family_histogram[WP_RES_FAMILY_COUNT];
} WpCorpusInventoryStats;

void wp_corpus_default_options(WpCorpusValidationOptions *options);
void wp_corpus_stats_clear(WpCorpusValidationStats *stats);
void wp_corpus_round_trip_stats_clear(WpCorpusRoundTripStats *stats);
void wp_corpus_inventory_stats_clear(WpCorpusInventoryStats *stats);

bool wp_corpus_validate_path(const char *path,
                             FILE *report,
                             const WpCorpusValidationOptions *options,
                             WpCorpusValidationStats *stats);

bool wp_corpus_round_trip_path(const char *path,
                               FILE *report,
                               WpCorpusRoundTripStats *stats);

bool wp_corpus_inventory_path(const char *path,
                              FILE *report,
                              WpCorpusInventoryStats *stats);

#endif
Commit message
This repository is read-only for this account.
Repository snapshot

Current branch
main
Visibility
public
Your access
Read
Remote
None
File activity

View file history