path: root/src/FileProcessor/FileProcessor.c

#define _POSIX_C_SOURCE 200809L
#define _XOPEN_SOURCE 700

#include "FileProcessor/FileProcessor.h"

#include "Logger/Logger.h"
#include "LuaEngine/LuaEngine.h"
#include "Path/Path.h"

#include <dirent.h>
#include <errno.h>
#include <fcntl.h>
#include <libxml/HTMLparser.h>
#include <libxml/HTMLtree.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <unistd.h>

#define COPY_BUFFER_SIZE 16384

static int has_html_extension(const char *path) {
  const char *extension = strrchr(path, '.');

  return extension != NULL && (strcasecmp(extension, ".html") == 0 ||
                               strcasecmp(extension, ".htm") == 0);
}

static void report_copy_error(const char *input_path, const char *output_path,
                              int error_number) {
  logger_error("cannot copy %s to %s: %s", input_path, output_path,
               strerror(error_number));
}

static int write_all(int file, const char *buffer, size_t length) {
  size_t offset = 0;

  while (offset < length) {
    ssize_t written = write(file, buffer + offset, length - offset);

    if (written > 0) {
      offset += (size_t)written;
    } else if (written == 0) {
      errno = EIO;
      return -1;
    } else if (errno != EINTR) {
      return -1;
    }
  }
  return 0;
}

static int copy_file_data(int input, int output) {
  char buffer[COPY_BUFFER_SIZE];

  for (;;) {
    ssize_t count = read(input, buffer, sizeof(buffer));

    if (count > 0 && write_all(output, buffer, (size_t)count) != 0) {
      return -1;
    }
    if (count == 0) {
      return 0;
    }
    if (count < 0 && errno != EINTR) {
      return -1;
    }
  }
}

static int copy_regular_file(const char *input_path, const char *output_path,
                             mode_t mode) {
  int input = open(input_path, O_RDONLY);
  int output;
  int result;
  int error_number;

  logger_debug("copying file %s -> %s", input_path, output_path);

  if (input < 0) {
    report_copy_error(input_path, output_path, errno);
    return -1;
  }

  output = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, mode & 0777);
  if (output < 0) {
    error_number = errno;
    close(input);
    report_copy_error(input_path, output_path, error_number);
    return -1;
  }

  result = copy_file_data(input, output);
  error_number = result == 0 ? 0 : errno;

  if (close(output) != 0 && result == 0) {
    result = -1;
    error_number = errno;
  }
  if (close(input) != 0 && result == 0) {
    result = -1;
    error_number = errno;
  }

  if (result != 0) {
    report_copy_error(input_path, output_path, error_number);
  }
  return result;
}

int process_file(const char *input_path, const char *output_path) {
  const int parse_options = HTML_PARSE_RECOVER | HTML_PARSE_NONET |
                            HTML_PARSE_NOERROR | HTML_PARSE_NOWARNING;
  struct stat status;
  htmlDocPtr document;
  int saved;

  if (stat(input_path, &status) != 0) {
    logger_error("cannot inspect %s: %s", input_path, strerror(errno));
    return -1;
  }
  if (!S_ISREG(status.st_mode)) {
    logger_error("input is not a regular file: %s", input_path);
    return -1;
  }

  logger_info("processing HTML %s -> %s", input_path, output_path);
  document = htmlReadFile(input_path, NULL, parse_options);
  if (document == NULL) {
    logger_error("unable to parse HTML file: %s", input_path);
    return -1;
  }

  if (process_html_document(document, input_path) != 0) {
    xmlFreeDoc(document);
    return -1;
  }

  saved = htmlSaveFileFormat(output_path, document, "UTF-8", 1);
  xmlFreeDoc(document);
  if (saved < 0) {
    logger_error("unable to write HTML file: %s", output_path);
    return -1;
  }
  return 0;
}

static int copy_symlink(const char *input_path, const char *output_path,
                        const struct stat *status) {
  size_t capacity = status->st_size > 0 ? (size_t)status->st_size + 1 : 256;
  char *target = malloc(capacity + 1);
  ssize_t length = -1;

  logger_debug("copying symbolic link %s -> %s", input_path, output_path);

  if (target == NULL) {
    logger_error("unable to allocate symbolic-link buffer");
    return -1;
  }

  while (length < 0) {
    ssize_t read_length = readlink(input_path, target, capacity);

    if (read_length < 0) {
      logger_error("cannot read symbolic link %s: %s", input_path,
                   strerror(errno));
      free(target);
      return -1;
    }
    if ((size_t)read_length < capacity) {
      length = read_length;
    } else {
      size_t larger_capacity;
      char *larger_target;

      if (capacity > (SIZE_MAX - 1) / 2) {
        logger_error("unable to allocate symbolic-link buffer");
        free(target);
        return -1;
      }
      larger_capacity = capacity * 2;
      larger_target = realloc(target, larger_capacity + 1);
      if (larger_target == NULL) {
        logger_error("unable to allocate symbolic-link buffer");
        free(target);
        return -1;
      }
      target = larger_target;
      capacity = larger_capacity;
    }
  }

  target[length] = '\0';
  if (symlink(target, output_path) != 0) {
    logger_error("cannot copy symbolic link %s: %s", input_path,
                 strerror(errno));
    free(target);
    return -1;
  }

  free(target);
  return 0;
}

static int copy_tree(const char *input_path, const char *output_path) {
  struct stat status;
  DIR *directory;
  int result = 0;

  if (lstat(input_path, &status) != 0) {
    logger_error("cannot inspect %s: %s", input_path, strerror(errno));
    return -1;
  }
  if (S_ISLNK(status.st_mode)) {
    return copy_symlink(input_path, output_path, &status);
  }
  if (S_ISREG(status.st_mode)) {
    return has_html_extension(input_path)
               ? process_file(input_path, output_path)
               : copy_regular_file(input_path, output_path, status.st_mode);
  }
  if (!S_ISDIR(status.st_mode)) {
    logger_error("unsupported file type: %s", input_path);
    return -1;
  }

  if (mkdir(output_path, status.st_mode & 0777) != 0) {
    logger_error("cannot create directory %s: %s", output_path,
                 strerror(errno));
    return -1;
  }

  directory = opendir(input_path);
  if (directory == NULL) {
    logger_error("cannot open directory %s: %s", input_path, strerror(errno));
    return -1;
  }

  while (result == 0) {
    struct dirent *entry;
    char *input_child;
    char *output_child;

    errno = 0;
    entry = readdir(directory);
    if (entry == NULL) {
      if (errno != 0) {
        logger_error("cannot read directory %s: %s", input_path,
                     strerror(errno));
        result = -1;
      }
      break;
    }
    if (strcmp(entry->d_name, ".") == 0 || strcmp(entry->d_name, "..") == 0) {
      continue;
    }

    input_child = path_join(input_path, entry->d_name);
    output_child = path_join(output_path, entry->d_name);
    if (input_child == NULL || output_child == NULL) {
      logger_error("unable to allocate child path");
      result = -1;
    } else {
      result = copy_tree(input_child, output_child);
    }
    free(input_child);
    free(output_child);
  }

  if (closedir(directory) != 0 && result == 0) {
    logger_error("cannot close directory %s: %s", input_path, strerror(errno));
    result = -1;
  }
  return result;
}

static int remove_tree(const char *path) {
  struct stat status;
  DIR *directory;
  int result = 0;

  if (lstat(path, &status) != 0) {
    if (errno == ENOENT) {
      return 0;
    }
    logger_error("cannot inspect output %s: %s", path, strerror(errno));
    return -1;
  }

  if (!S_ISDIR(status.st_mode) || S_ISLNK(status.st_mode)) {
    if (unlink(path) != 0) {
      logger_error("cannot remove output %s: %s", path, strerror(errno));
      return -1;
    }
    return 0;
  }

  directory = opendir(path);
  if (directory == NULL) {
    logger_error("cannot open output directory %s: %s", path, strerror(errno));
    return -1;
  }

  while (result == 0) {
    struct dirent *entry;
    char *child;

    errno = 0;
    entry = readdir(directory);
    if (entry == NULL) {
      if (errno != 0) {
        logger_error("cannot read output directory %s: %s", path,
                     strerror(errno));
        result = -1;
      }
      break;
    }
    if (strcmp(entry->d_name, ".") == 0 || strcmp(entry->d_name, "..") == 0) {
      continue;
    }

    child = path_join(path, entry->d_name);
    if (child == NULL) {
      logger_error("unable to allocate output child path");
      result = -1;
    } else {
      result = remove_tree(child);
    }
    free(child);
  }

  if (closedir(directory) != 0 && result == 0) {
    logger_error("cannot close output directory %s: %s", path, strerror(errno));
    result = -1;
  }
  if (result == 0 && rmdir(path) != 0) {
    logger_error("cannot remove output directory %s: %s", path,
                 strerror(errno));
    result = -1;
  }
  return result;
}

int process_directory(const char *input_path, const char *output_path,
                      int overwrite_output) {
  struct stat status;
  struct stat output_status;
  char *resolved_input;
  char *resolved_output;
  int result;

  if (stat(input_path, &status) != 0) {
    logger_error("cannot inspect %s: %s", input_path, strerror(errno));
    return -1;
  }
  if (!S_ISDIR(status.st_mode)) {
    logger_error("input is not a directory: %s", input_path);
    return -1;
  }

  resolved_input = realpath(input_path, NULL);
  if (lstat(output_path, &output_status) == 0 &&
      !S_ISLNK(output_status.st_mode)) {
    resolved_output = realpath(output_path, NULL);
  } else {
    resolved_output = path_resolve_new(output_path);
  }
  if (resolved_input == NULL || resolved_output == NULL) {
    logger_error("cannot resolve input or output path");
    free(resolved_input);
    free(resolved_output);
    return -1;
  }

  if (path_is_within(resolved_input, resolved_output)) {
    logger_error("output directory cannot be inside input: %s", output_path);
    result = -1;
  } else if (overwrite_output &&
             path_is_within(resolved_output, resolved_input)) {
    logger_error("output directory cannot contain input: %s", output_path);
    result = -1;
  } else if (overwrite_output && remove_tree(output_path) != 0) {
    result = -1;
  } else {
    logger_info("processing directory %s -> %s", input_path, output_path);
    result = copy_tree(input_path, output_path);
  }

  free(resolved_input);
  free(resolved_output);
  return result;
}