[gnome-maps/wip/mlundblad/transit-plugin-gtfs-local: 5/6] WIP: Add class for parsing GTFS files

[Marcus Lundblad <mlundblad src gnome org>]

commit 4903000102c87c5d614b1a492ddcd1e8c5e35889
Author: Marcus Lundblad <ml update uu se>
Date:   Sat Feb 15 09:29:35 2020 +0100

    WIP: Add class for parsing GTFS files

 lib/maps-gtfs.c | 559 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 lib/maps-gtfs.h |  46 +++++
 lib/meson.build |   2 +
 3 files changed, 607 insertions(+)
---
diff --git a/lib/maps-gtfs.c b/lib/maps-gtfs.c
new file mode 100644
index 00000000..11253af9
--- /dev/null
+++ b/lib/maps-gtfs.c
@@ -0,0 +1,559 @@
+/*
+ * Copyright (c) 2020 Marcus Lundblad
+ *
+ * GNOME Maps is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2 of the License, or (at your
+ * option) any later version.
+ *
+ * GNOME Maps is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+ * or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with GNOME Maps; if not, see <http://www.gnu.org/licenses/>
+ *
+ * Author: Marcus Lundblad <ml update uu se>
+ */
+
+#include "maps-gtfs.h"
+#include "maps-gtfs-agency.h"
+#include "mapsintl.h"
+
+#include <archive.h>
+#include <archive_entry.h>
+#include <glib.h>
+#include <string.h>
+
+#define MAPS_GTFS_ERROR maps_gtfs_error_quark ()
+
+static const int BLOCK_SIZE = 10 * 1024;
+
+GQuark
+maps_gtfs_error_quark (void)
+{
+  return g_quark_from_static_string ("maps-gtfs-error");
+}
+
+struct _MapsGTFSPrivate
+{
+  gchar *file;
+  gchar *read_buffer;
+  guint read_size;
+  guint read_pos;
+  GTimeZone *timezone;
+  GHashTable *agencies;
+};
+
+enum {
+  PROP_0,
+};
+
+/*
+ * File types expected in a GTFS feed.
+ * https://developers.google.com/transit/gtfs/reference/#dataset_files
+ */
+typedef enum {
+  AGENCY,
+  STOPS,
+  ROUTES,
+  TRIPS,
+  STOP_TIMES,
+  CALENDAR,
+  CALENDAR_DATES,
+  FARE_ATTRIBUTES,
+  FARE_RULES,
+  SHAPES,
+  FREQUENCIES,
+  TRANSFERS,
+  PATHWAYS,
+  LEVELS,
+  FEED_INFO,
+  TRANSLATIONS,
+  ATTRIBUTIONS
+} FileType;
+
+G_DEFINE_TYPE_WITH_PRIVATE (MapsGTFS, maps_gtfs, G_TYPE_OBJECT)
+
+static void
+maps_gtfs_dispose (GObject *object)
+{
+  MapsGTFS *gtfs = MAPS_GTFS (object);
+  MapsGTFSPrivate *priv = maps_gtfs_get_instance_private (gtfs);
+
+  g_clear_pointer (&priv->file, g_free);
+  g_clear_pointer (&priv->read_buffer, g_free);
+  g_clear_pointer (&priv->agencies, g_hash_table_destroy);
+  g_clear_pointer (&priv->timezone, g_time_zone_unref);
+
+  G_OBJECT_CLASS (maps_gtfs_parent_class)->dispose (object);
+}
+
+static void
+maps_gtfs_class_init (MapsGTFSClass *klass)
+{
+  GObjectClass *maps_class = G_OBJECT_CLASS (klass);
+
+  maps_class->dispose = maps_gtfs_dispose;
+}
+
+static void
+maps_gtfs_init (MapsGTFS *gtfs)
+{
+  gtfs->priv = maps_gtfs_get_instance_private (gtfs);
+  gtfs->priv->file = NULL;
+  gtfs->priv->read_buffer = g_malloc (BLOCK_SIZE);
+  /* indicate we need to read a new chunk of data when reading the first line */
+  gtfs->priv->read_pos = 0;
+  gtfs->priv->read_size = 0;
+  gtfs->priv->agencies =
+    g_hash_table_new_full (g_str_hash, g_str_equal, g_free, g_object_unref);
+}
+
+MapsGTFS *
+maps_gtfs_new (char *file)
+{
+  MapsGTFS *gtfs = g_object_new (MAPS_TYPE_GTFS, NULL);
+  MapsGTFSPrivate *priv = maps_gtfs_get_instance_private (gtfs);
+
+  priv->file = g_strdup (file);
+  return gtfs;
+}
+
+static void
+reset_read_buffer (MapsGTFS *gtfs)
+{
+  MapsGTFSPrivate *priv = maps_gtfs_get_instance_private (gtfs);
+
+  priv->read_pos = 0;
+  priv->read_size = 0;
+}
+
+static char *
+read_line (MapsGTFS *gtfs, struct archive *a)
+{
+  MapsGTFSPrivate *priv = maps_gtfs_get_instance_private (gtfs);
+  int i;
+  gchar *result;
+
+  if (priv->read_pos == priv->read_size) {
+    g_debug ("Read new chunk");
+    priv->read_size = archive_read_data (a, priv->read_buffer, BLOCK_SIZE);
+    priv->read_pos = 0;
+
+    if (priv->read_size < 0) {
+      g_error ("Failed to read data");
+      return NULL;
+    } else if (priv->read_size == 0) {
+      g_debug ("No more data to read");
+      return NULL;
+    }
+  }
+
+  g_debug ("Read pos at start: %d", priv->read_pos);
+
+  /* find next line break */
+  i = priv->read_pos;
+  while (priv->read_buffer[i] != '\n' && priv->read_buffer[i] != '\r' &&
+         i < priv->read_size) {
+    i++;
+  }
+
+  /* if we didn't read til the end of the buffer, just copy the line */
+  if (i < priv->read_size) {
+    result = g_strndup (priv->read_buffer + priv->read_pos, i - priv->read_pos);
+    priv->read_pos = i;
+    /* skip upcoming newline(s) */
+    while (priv->read_buffer[priv->read_pos] == '\n' ||
+           priv->read_buffer[priv->read_pos] == '\r') {
+      priv->read_pos++;
+    }
+  } else {
+    /* copy the first part up to the buffer end */
+    g_autofree gchar *first =
+      g_strndup (priv->read_buffer + priv->read_pos, i - priv->read_pos);
+
+    /* read next chunk */
+    g_debug ("Read next chunk");
+    priv->read_size = archive_read_data (a, priv->read_buffer, BLOCK_SIZE);
+    priv->read_pos = 0;
+
+    if (priv->read_size < 0) {
+      g_error ("Failed to read data");
+      return NULL;
+    } else if (priv->read_size == 0) {
+      g_debug ("No more data to read");
+      return NULL;
+    }
+
+    /* find next line break */
+    i = priv->read_pos;
+    while (priv->read_buffer[i] != '\n' && priv->read_buffer[i] != '\r' &&
+           i < priv->read_size) {
+      i++;
+    }
+    priv->read_pos = i;
+
+    g_debug ("Found next line break at position: %d", i);
+
+    /* copy next part, read from new buffer */
+    g_autofree gchar *second = g_strndup (priv->read_buffer, i);
+
+    /* skip upcoming newline(s) */
+    while (priv->read_buffer[priv->read_pos] == '\n' ||
+           priv->read_buffer[priv->read_pos] == '\r') {
+      priv->read_pos++;
+    }
+
+    /* concatenate parts */
+    result = g_strdup_printf("%s%s", first, second);
+  }
+
+  return result;
+}
+
+static gchar **
+parse_header (const gchar *filename, const gchar *header, ...)
+{
+  gchar **headers = g_strsplit (header, ",", -1);
+  va_list ap;
+  gboolean missing_required = FALSE;
+  g_autoptr(GHashTable) fields = g_hash_table_new (g_str_hash, g_str_equal);
+
+  va_start (ap, header);
+  for (;;) {
+    gchar *field = va_arg (ap, gchar *);
+    gboolean required;
+
+    if (!field)
+      break;
+
+    required = va_arg (ap, gboolean);
+    g_hash_table_insert (fields, field, field);
+
+    if (required) {
+      gboolean found_required = FALSE;
+
+      for (int i = 0; headers[i]; i++) {
+        if (g_str_equal (headers[i], field))
+          found_required = TRUE;
+      }
+
+      if (!found_required) {
+        missing_required = TRUE;
+        g_error ("Missing required field %s in %s", field, filename);
+        break;
+      }
+    }
+  }
+  va_end (ap);
+
+  for (int i = 0; headers[i]; i++) {
+    const gchar *found_field = g_hash_table_lookup (fields, headers[i]);
+
+    if (!found_field)
+      g_warning ("Unknown field %s in %s", headers[i], filename);
+  }
+
+  g_debug ("missing required: %s", missing_required ? "true" : "false");
+
+  if (missing_required) {
+    g_strfreev (headers);
+    return NULL;
+  } else {
+    return headers;
+  }
+}
+
+static GHashTable *
+parse_line (const gchar *line, gchar **header_fields, gint num_fields)
+{
+  gchar **fields = NULL;
+  gboolean in_quotes = FALSE;
+  gboolean escape = FALSE;
+
+  if (line[0] == '\0')
+    return NULL;
+
+  if (g_strstr_len (line, -1, "\"")) {
+    // if the input line contains no quoted strings, just do a simple split
+    fields = g_strsplit (line, ",", -1);
+  } else {
+    g_autofree gchar *field = g_malloc0 (strlen (line) + 1);
+    gint index = 0;
+    gint j = 0;
+
+    fields = g_malloc0_n (num_fields + 1, sizeof (gchar *));
+    for (gint i = 0; line[i]; i++) {
+      gchar c = line[i];
+
+      if (index >= num_fields) {
+        g_error ("Too many fields read");
+        g_strfreev (fields);
+        return NULL;
+      }
+
+      if (escape) {
+        field[j] = c;
+        escape = FALSE;
+        j++;
+      } else {
+        escape = (c == '\\');
+      }
+
+      if (c == ',' && !in_quotes) {
+        fields[index] = g_strdup (field);
+        for (gint k = 0; k <= num_fields; k++)
+          field[k] = '\0';
+        index++;
+        continue;
+      }
+
+      if (c == '"') {
+        in_quotes = !in_quotes;
+      } else {
+        field[j] = c;
+        j++;
+      }
+    }
+
+    fields[index] = g_strdup (field);
+  }
+
+  if (g_strv_length (fields) != num_fields) {
+    g_error ("Number of fields read didn't match header");
+    g_strfreev (fields);
+    return NULL;
+  } else {
+    GHashTable *result =
+      g_hash_table_new_full (g_str_hash, g_str_equal, NULL, g_free);
+
+    for (gint i = 0; i < num_fields; i++) {
+      g_hash_table_insert (result, header_fields[i], fields[i]);
+    }
+    return result;
+  }
+}
+
+static gboolean
+process_agency (MapsGTFS *gtfs, GHashTable *entry)
+{
+  MapsGTFSPrivate *priv = maps_gtfs_get_instance_private (gtfs);
+
+  MapsGTFSAgency *agency;
+  GTimeZone *tz;
+
+  gchar *id = g_hash_table_lookup (entry, "agency_id");
+  gchar *name = g_hash_table_lookup (entry, "agency_name");
+  gchar *url = g_hash_table_lookup (entry, "agency_name");
+  gchar *timezone = g_hash_table_lookup (entry, "agency_timezone");
+  gchar *lang = g_hash_table_lookup (entry, "agency_lang");
+  gchar *phone = g_hash_table_lookup (entry, "agency_phone");
+  gchar *fare_url = g_hash_table_lookup (entry, "agency_fare_url");
+  gchar *email = g_hash_table_lookup (entry, "agency_email");
+
+  /* agency_id is required if there's more than one agency */
+  if (!id && g_hash_table_size (priv->agencies) > 0) {
+    g_error ("Field agency_id is required when there's more than one agency");
+    return FALSE;
+  }
+
+  tz = g_time_zone_new (timezone);
+
+  if (!priv->timezone) {
+    priv->timezone = tz;
+  } else if (!g_str_equal (g_time_zone_get_identifier (tz),
+                           g_time_zone_get_identifier (priv->timezone))) {
+    g_error ("All agencies in the feed must use the timezone");
+    g_time_zone_unref (tz);
+    return FALSE;
+  }
+
+  agency =
+    maps_gtfs_agency_new (id, name, url, tz, lang, phone, fare_url, email);
+  g_hash_table_insert (priv->agencies, g_strdup (id), agency);
+
+  return TRUE;
+}
+
+static gboolean
+process_stop (MapsGTFS *gtfs, GHashTable *entry)
+{
+
+  return TRUE;
+}
+
+static gboolean
+process_entry (MapsGTFS *gtfs, FileType file_type, GHashTable *entry)
+{
+  switch (file_type) {
+  case AGENCY: return process_agency (gtfs, entry);
+  case STOPS:  return process_stop (gtfs, entry);
+  default:     return TRUE; /* this shouldn't happen */
+  }
+}
+
+static gboolean
+parse_file (MapsGTFS *gtfs, struct archive *a, FileType file_type,
+            const gchar *filename)
+{
+  g_autofree gchar *header;
+  g_auto(GStrv) header_fields = NULL;
+  gint num_fields;
+
+  reset_read_buffer (gtfs);
+  header = read_line (gtfs, a);
+
+  if (!header) {
+    g_error ("Failed to read header in %s", filename);
+    return FALSE;
+  }
+
+  g_debug ("Header for %s: %s", filename, header);
+
+  switch (file_type) {
+  case AGENCY:
+    header_fields = parse_header (filename, header, "agency_id", FALSE,
+                                  "agency_name", TRUE, "agency_url", TRUE,
+                                  "agency_timezone", TRUE, "agency_lang", FALSE,
+                                  "agency_phone", FALSE,
+                                  "agency_fare_url", FALSE,
+                                  "agency_email", FALSE, NULL);
+    break;
+  case STOPS:
+    header_fields = parse_header (filename, header, "stop_id", TRUE,
+                                  "stop_code", FALSE, "stop_name", FALSE,
+                                  "stop_desc", FALSE, "stop_lat", FALSE,
+                                  "stop_lon", FALSE, "zone_id", FALSE,
+                                  "stop_url", FALSE, "location_type", FALSE,
+                                  "parent_station", FALSE, "stop_timezone", FALSE,
+                                  "wheelchair_boarding", FALSE, "level_id", FALSE,
+                                  "platform_code", FALSE, NULL);
+    break;
+  default:
+    // skip other files for now
+    return TRUE;
+    break;
+  }
+
+  num_fields = g_strv_length (header_fields);
+
+  for (;;) {
+    g_autofree gchar *line = read_line (gtfs, a);
+    g_autoptr(GHashTable) fields = NULL;
+    if (!line)
+      break;
+    g_debug ("Line: %s", line);
+    fields = parse_line (line, header_fields, num_fields);
+    if (!process_entry (gtfs, file_type, fields)) {
+      g_error ("Failed to parse entry in %s: %s", filename, line);
+      return FALSE;
+    }
+  }
+
+  return TRUE;
+}
+
+static gboolean
+parse_entries (MapsGTFS *gtfs, struct archive *a)
+{
+  struct archive_entry *entry;
+  gboolean found_agency = FALSE;
+  gboolean found_stops = FALSE;
+  gboolean found_routes = FALSE;
+  gboolean found_trips = FALSE;
+  gboolean found_stop_times = FALSE;
+
+  while (archive_read_next_header(a, &entry) == ARCHIVE_OK) {
+    const char *filename = archive_entry_pathname (entry);
+    FileType file_type;
+
+    g_debug ("Entry %s", archive_entry_pathname(entry));
+
+    if (g_str_equal (filename, "agency.txt")) {
+      file_type = AGENCY;
+      found_agency = TRUE;
+    } else if (g_str_equal (filename, "stops.txt")) {
+      file_type = STOPS;
+      found_stops = TRUE;
+    } else if (g_str_equal (filename, "routes.txt")) {
+      file_type = ROUTES;
+      found_routes = TRUE;
+    } else if (g_str_equal (filename, "trips.txt")) {
+      file_type = TRIPS;
+      found_trips = TRUE;
+    } else if (g_str_equal (filename, "stop_times.txt")) {
+      file_type = STOP_TIMES;
+      found_stop_times = TRUE;
+    } else if (g_str_equal (filename, "calendar.txt")) {
+      file_type = CALENDAR;
+    } else if (g_str_equal (filename, "calendar_dates.txt")) {
+      file_type = CALENDAR_DATES;
+    } else if (g_str_equal (filename, "fare_attributes.txt")) {
+      file_type = FARE_ATTRIBUTES;
+    } else if (g_str_equal (filename, "fare_rules.txt")) {
+      file_type = FARE_RULES;
+    } else if (g_str_equal (filename, "shapes.txt")) {
+      file_type = SHAPES;
+    } else if (g_str_equal (filename, "frequencies.txt")) {
+      file_type = FREQUENCIES;
+    } else if (g_str_equal (filename, "transfers.txt")) {
+      file_type = TRANSFERS;
+    } else if (g_str_equal (filename, "pathways.txt")) {
+      file_type = PATHWAYS;
+    } else if (g_str_equal (filename, "levels.txt")) {
+      file_type = LEVELS;
+    } else if (g_str_equal (filename, "feed_info.txt")) {
+      file_type = FEED_INFO;
+    } else if (g_str_equal (filename, "translations.txt")) {
+      file_type = TRANSLATIONS;
+    } else if (g_str_equal (filename, "attributions.txt")) {
+      file_type = ATTRIBUTIONS;
+    } else {
+      g_warning ("Unrecognized file: %s", filename);
+      continue;
+    }
+
+    if (!parse_file (gtfs, a, file_type, filename))
+        return FALSE;
+
+    archive_read_data_skip (a);
+  }
+
+  if (found_agency && found_stops && found_routes && found_trips && found_stop_times)
+    return TRUE;
+  else
+    return FALSE;
+}
+
+void
+maps_gtfs_parse (MapsGTFS *gtfs, GError **error)
+{
+  MapsGTFSPrivate *priv = maps_gtfs_get_instance_private (gtfs);
+  struct archive *a;
+  int r;
+
+  a = archive_read_new ();
+  archive_read_support_format_zip (a);
+  r = archive_read_open_filename (a, priv->file, BLOCK_SIZE);
+
+  if (r != ARCHIVE_OK)
+    {
+      *error = g_error_new_literal (MAPS_GTFS_ERROR, 0,
+                                    _("Failed to open GTFS ZIP file"));
+      return;
+    }
+
+  if (!parse_entries (gtfs, a)) {
+    *error = g_error_new_literal (MAPS_GTFS_ERROR, 0,
+                                  _("Failed to parse GTFS file"));
+    return;
+  }
+
+  r = archive_read_free(a);
+  if (r != ARCHIVE_OK)
+    *error = g_error_new_literal (MAPS_GTFS_ERROR, 0,
+                                  _("Failed to free GTFS ZIP file"));
+
+}
diff --git a/lib/maps-gtfs.h b/lib/maps-gtfs.h
new file mode 100644
index 00000000..c1417eb9
--- /dev/null
+++ b/lib/maps-gtfs.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2020 Marcus Lundblad
+ *
+ * GNOME Maps is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2 of the License, or (at your
+ * option) any later version.
+ *
+ * GNOME Maps is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+ * or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with GNOME Maps; if not, see <http://www.gnu.org/licenses/>
+ *
+ * Author: Marcus Lundblad <ml update uu se>
+ */
+
+#ifndef __MAPS_GTFS__
+#define __MAPS_GTFS__
+
+#include <glib-object.h>
+
+#define MAPS_TYPE_GTFS maps_gtfs_get_type ()
+G_DECLARE_FINAL_TYPE(MapsGTFS, maps_gtfs, MAPS, GTFS, GObject)
+
+typedef struct _MapsGTFSPrivate MapsGTFSPrivate;
+
+struct _MapsGTFS
+{
+  GObject parent_instance;
+  MapsGTFSPrivate *priv;
+};
+
+
+struct _MapsGTFSClass
+{
+  GObjectClass parent_class;
+};
+
+MapsGTFS *maps_gtfs_new (char *file);
+
+void maps_gtfs_parse (MapsGTFS *gtfs, GError **error);
+
+#endif //__MAPS_GTFS__
diff --git a/lib/meson.build b/lib/meson.build
index 170d43d9..65dfa193 100644
--- a/lib/meson.build
+++ b/lib/meson.build
@@ -4,6 +4,7 @@ headers_private = files(
        'maps-contact.h',
        'maps-contact-store.h',
        'maps-file-tile-source.h',
+       'maps-gtfs.h',
        'maps-gtfs-agency.h',
        'maps-gtfs-stop.h',
        'maps-osm.h',
@@ -19,6 +20,7 @@ sources = files(
        'maps-contact.c',
        'maps-contact-store.c',
        'maps-file-tile-source.c',
+       'maps-gtfs.c',
        'maps-gtfs-agency.c',
        'maps-gtfs-stop.c',
        'maps-osm.c',