Refactore parse-xml.c into parse.c and parse-xml.c

This should help us to move parsing that is not XML related to other
files, hopefully making the code cleaner.

Signed-off-by: Miika Turkia <miika.turkia@gmail.com>

7 files changed, 550 insertions, 446 deletions

diff --git a/core/CMakeLists.txt b/core/CMakeLists.txt
index f063f3aa5..31b82ebbb 100644
--- a/core/CMakeLists.txt
+++ b/core/CMakeLists.txt
@@ -50,6 +50,7 @@ set(SUBSURFACE_CORE_LIB_SRCS
 	membuffer.c
 	ostctools.c
 	parse-xml.c
+	parse.c
 	planner.c
 	plannernotes.c
 	profile.c
diff --git a/core/membuffer.h b/core/membuffer.h
index fd59d7140..bd0a46d3a 100644
--- a/core/membuffer.h
+++ b/core/membuffer.h
@@ -7,6 +7,7 @@ extern "C" {
 #endif
 
 #include <ctype.h>
+#include "units.h"
 
 struct membuffer {
 	unsigned int len, alloc;
diff --git a/core/parse-xml.c b/core/parse-xml.c
index 6fd298710..4de4ab867 100644
--- a/core/parse-xml.c
+++ b/core/parse-xml.c
@@ -22,6 +22,7 @@
 #include "gettext.h"
 
 #include "dive.h"
+#include "parse.h"
 #include "divelist.h"
 #include "device.h"
 #include "membuffer.h"
@@ -33,148 +34,10 @@ int diveid = -1;
 
 static xmlDoc *test_xslt_transforms(xmlDoc *doc, const char **params);
 
-/* the dive table holds the overall dive list; target table points at
- * the table we are currently filling */
-struct dive_table dive_table;
-struct dive_table *target_table = NULL;
-
-/* Trim a character string by removing leading and trailing white space characters.
- * Parameter: a pointer to a null-terminated character string (buffer);
- * Return value: length of the trimmed string, excluding the terminal 0x0 byte
- * The original pointer (buffer) remains valid after this function has been called
- * and points to the trimmed string */
-int trimspace(char *buffer)
-{
-	int i, size, start, end;
-	size = strlen(buffer);
-
-	if (!size)
-		return 0;
-	for(start = 0; isspace(buffer[start]); start++)
-		if (start >= size) return 0;	// Find 1st character following leading whitespace
-	for(end = size - 1; isspace(buffer[end]); end--) // Find last character before trailing whitespace
-		if (end <= 0) return 0;
-	for(i = start; i <= end; i++)		// Move the nonspace characters to the start of the string
-		buffer[i-start] = buffer[i];
-	size = end - start + 1;
-	buffer[size] = 0x0;			// then terminate the string
-	return size;				// return string length
-}
-
-/*
- * Clear a dive_table
- */
-void clear_table(struct dive_table *table)
-{
-	for (int i = 0; i < table->nr; i++)
-		free(table->dives[i]);
-	table->nr = 0;
-}
-
-/*
- * Add a dive into the dive_table array
- */
-void record_dive_to_table(struct dive *dive, struct dive_table *table)
-{
-	assert(table != NULL);
-	struct dive **dives = grow_dive_table(table);
-	int nr = table->nr;
-
-	dives[nr] = fixup_dive(dive);
-	table->nr = nr + 1;
-}
-
-void record_dive(struct dive *dive)
-{
-	record_dive_to_table(dive, &dive_table);
-}
-
-static void start_match(const char *type, const char *name, char *buffer)
-{
-	if (verbose > 2)
-		printf("Matching %s '%s' (%s)\n",
-		       type, name, buffer);
-}
-
-static void nonmatch(const char *type, const char *name, char *buffer)
-{
-	if (verbose > 1)
-		printf("Unable to match %s '%s' (%s)\n",
-		       type, name, buffer);
-}
-
-typedef void (*matchfn_t)(char *buffer, void *);
-
-static int match(const char *pattern, int plen,
-		 const char *name,
-		 matchfn_t fn, char *buf, void *data)
-{
-	switch (name[plen]) {
-	case '\0':
-	case '.':
-		break;
-	default:
-		return 0;
-	}
-	if (memcmp(pattern, name, plen))
-		return 0;
-	fn(buf, data);
-	return 1;
-}
-
-
 struct units xml_parsing_units;
 const struct units SI_units = SI_UNITS;
 const struct units IMPERIAL_units = IMPERIAL_UNITS;
 
-/*
- * Dive info as it is being built up..
- */
-#define MAX_EVENT_NAME 128
-static struct divecomputer *cur_dc;
-static struct dive *cur_dive;
-static struct dive_site *cur_dive_site;
-degrees_t cur_latitude, cur_longitude;
-static dive_trip_t *cur_trip = NULL;
-static struct sample *cur_sample;
-static struct picture *cur_picture;
-static union {
-	struct event event;
-	char allocation[sizeof(struct event)+MAX_EVENT_NAME];
-} event_allocation = { .event.deleted = 1 };
-#define cur_event event_allocation.event
-static struct {
-	struct {
-		const char *model;
-		uint32_t deviceid;
-		const char *nickname, *serial_nr, *firmware;
-	} dc;
-} cur_settings;
-static bool in_settings = false;
-static bool in_userid = false;
-static struct tm cur_tm;
-static int cur_cylinder_index, cur_ws_index;
-static int lastcylinderindex, next_o2_sensor;
-static int o2pressure_sensor;
-static struct extra_data cur_extra_data;
-
-/*
- * If we don't have an explicit dive computer,
- * we use the implicit one that every dive has..
- */
-static struct divecomputer *get_dc(void)
-{
-	return cur_dc ?: &cur_dive->dc;
-}
-
-static enum import_source {
-	UNKNOWN,
-	LIBDIVECOMPUTER,
-	DIVINGLOG,
-	UDDF,
-	SSRF_WS,
-} import_source;
-
 static void divedate(const char *buffer, timestamp_t *when)
 {
 	int d, m, y;
@@ -561,15 +424,6 @@ static void cylindersize(char *buffer, volume_t *volume)
 	}
 }
 
-static void utf8_string(char *buffer, void *_res)
-{
-	char **res = _res;
-	int size;
-	size = trimspace(buffer);
-	if(size)
-		*res = strdup(buffer);
-}
-
 static void event_name(char *buffer, char *name)
 {
 	int size = trimspace(buffer);
@@ -1537,305 +1391,6 @@ static void try_to_fill_dive_site(struct dive_site **ds_p, const char *name, cha
 	nonmatch("divesite", name, buf);
 }
 
-/*
- * While in some formats file boundaries are dive boundaries, in many
- * others (as for example in our native format) there are
- * multiple dives per file, so there can be other events too that
- * trigger a "new dive" marker and you may get some nesting due
- * to that. Just ignore nesting levels.
- * On the flipside it is possible that we start an XML file that ends
- * up having no dives in it at all - don't create a bogus empty dive
- * for those. It's not entirely clear what is the minimum set of data
- * to make a dive valid, but if it has no location, no date and no
- * samples I'm pretty sure it's useless.
- */
-static bool is_dive(void)
-{
-	return (cur_dive &&
-		(cur_dive->dive_site_uuid || cur_dive->when || cur_dive->dc.samples));
-}
-
-static void reset_dc_info(struct divecomputer *dc)
-{
-	/* WARN: reset dc info does't touch the dc? */
-	(void) dc;
-	lastcylinderindex = 0;
-}
-
-static void reset_dc_settings(void)
-{
-	free((void *)cur_settings.dc.model);
-	free((void *)cur_settings.dc.nickname);
-	free((void *)cur_settings.dc.serial_nr);
-	free((void *)cur_settings.dc.firmware);
-	cur_settings.dc.model = NULL;
-	cur_settings.dc.nickname = NULL;
-	cur_settings.dc.serial_nr = NULL;
-	cur_settings.dc.firmware = NULL;
-	cur_settings.dc.deviceid = 0;
-}
-
-static void settings_start(void)
-{
-	in_settings = true;
-}
-
-static void settings_end(void)
-{
-	in_settings = false;
-}
-
-static void dc_settings_start(void)
-{
-	reset_dc_settings();
-}
-
-static void dc_settings_end(void)
-{
-	create_device_node(cur_settings.dc.model, cur_settings.dc.deviceid, cur_settings.dc.serial_nr,
-			   cur_settings.dc.firmware, cur_settings.dc.nickname);
-	reset_dc_settings();
-}
-
-static void dive_site_start(void)
-{
-	if (cur_dive_site)
-		return;
-	cur_dive_site = calloc(1, sizeof(struct dive_site));
-}
-
-static void dive_site_end(void)
-{
-	if (!cur_dive_site)
-		return;
-	if (cur_dive_site->taxonomy.nr == 0) {
-		free(cur_dive_site->taxonomy.category);
-		cur_dive_site->taxonomy.category = NULL;
-	}
-	if (cur_dive_site->uuid) {
-		struct dive_site *ds = alloc_or_get_dive_site(cur_dive_site->uuid);
-		merge_dive_site(ds, cur_dive_site);
-
-		if (verbose > 3)
-			printf("completed dive site uuid %x8 name {%s}\n", ds->uuid, ds->name);
-	}
-	free_taxonomy(&cur_dive_site->taxonomy);
-	free(cur_dive_site);
-	cur_dive_site = NULL;
-}
-
-// now we need to add the code to parse the parts of the divesite enry
-
-static void dive_start(void)
-{
-	if (cur_dive)
-		return;
-	cur_dive = alloc_dive();
-	reset_dc_info(&cur_dive->dc);
-	memset(&cur_tm, 0, sizeof(cur_tm));
-	if (cur_trip) {
-		add_dive_to_trip(cur_dive, cur_trip);
-		cur_dive->tripflag = IN_TRIP;
-	}
-	o2pressure_sensor = 1;
-}
-
-static void dive_end(void)
-{
-	if (!cur_dive)
-		return;
-	if (!is_dive())
-		free(cur_dive);
-	else
-		record_dive_to_table(cur_dive, target_table);
-	cur_dive = NULL;
-	cur_dc = NULL;
-	cur_latitude.udeg = 0;
-	cur_longitude.udeg = 0;
-	cur_cylinder_index = 0;
-	cur_ws_index = 0;
-}
-
-static void trip_start(void)
-{
-	if (cur_trip)
-		return;
-	dive_end();
-	cur_trip = calloc(1, sizeof(dive_trip_t));
-	memset(&cur_tm, 0, sizeof(cur_tm));
-}
-
-static void trip_end(void)
-{
-	if (!cur_trip)
-		return;
-	insert_trip(&cur_trip);
-	cur_trip = NULL;
-}
-
-static void event_start(void)
-{
-	memset(&cur_event, 0, sizeof(cur_event));
-	cur_event.deleted = 0;	/* Active */
-}
-
-static void event_end(void)
-{
-	struct divecomputer *dc = get_dc();
-	if (cur_event.type == 123) {
-		struct picture *pic = alloc_picture();
-		pic->filename = strdup(cur_event.name);
-		/* theoretically this could fail - but we didn't support multi year offsets */
-		pic->offset.seconds = cur_event.time.seconds;
-		dive_add_picture(cur_dive, pic);
-	} else {
-		struct event *ev;
-		/* At some point gas change events did not have any type. Thus we need to add
-		 * one on import, if we encounter the type one missing.
-		 */
-		if (cur_event.type == 0 && strcmp(cur_event.name, "gaschange") == 0)
-			cur_event.type = cur_event.value >> 16 > 0 ? SAMPLE_EVENT_GASCHANGE2 : SAMPLE_EVENT_GASCHANGE;
-		ev = add_event(dc, cur_event.time.seconds,
-			       cur_event.type, cur_event.flags,
-			       cur_event.value, cur_event.name);
-
-		/*
-		 * Older logs might mark the dive to be CCR by having an "SP change" event at time 0:00. Better
-		 * to mark them being CCR on import so no need for special treatments elsewhere on the code.
-		 */
-		if (ev && cur_event.time.seconds == 0 && cur_event.type == SAMPLE_EVENT_PO2 && cur_event.value && dc->divemode==OC) {
-			dc->divemode = CCR;
-		}
-
-		if (ev && event_is_gaschange(ev)) {
-			/* See try_to_fill_event() on why the filled-in index is one too big */
-			ev->gas.index = cur_event.gas.index-1;
-			if (cur_event.gas.mix.o2.permille || cur_event.gas.mix.he.permille)
-				ev->gas.mix = cur_event.gas.mix;
-		}
-	}
-	cur_event.deleted = 1;	/* No longer active */
-}
-
-static void picture_start(void)
-{
-	cur_picture = alloc_picture();
-}
-
-static void picture_end(void)
-{
-	dive_add_picture(cur_dive, cur_picture);
-	cur_picture = NULL;
-}
-
-static void cylinder_start(void)
-{
-}
-
-static void cylinder_end(void)
-{
-	cur_cylinder_index++;
-}
-
-static void ws_start(void)
-{
-}
-
-static void ws_end(void)
-{
-	cur_ws_index++;
-}
-
-/*
- * By default the sample data does not change unless the
- * save-file gives an explicit new value. So we copy the
- * data from the previous sample if one exists, and then
- * the parsing will update it as necessary.
- *
- * There are a few exceptions, like the sample pressure:
- * missing sample pressure doesn't mean "same as last
- * time", but "interpolate". We clear those ones
- * explicitly.
- *
- * NOTE! We default sensor use to 0, 1 respetively for
- * the two sensors, but for CCR dives with explicit
- * OXYGEN bottles we set the secondary sensor to that.
- * Then the primary sensor will be either the first
- * or the second cylinder depending on what isn't an
- * oxygen cylinder.
- */
-static void sample_start(void)
-{
-	struct divecomputer *dc = get_dc();
-	struct sample *sample = prepare_sample(dc);
-
-	if (sample != dc->sample) {
-		memcpy(sample, sample-1, sizeof(struct sample));
-		sample->pressure[0].mbar = 0;
-		sample->pressure[1].mbar = 0;
-	} else {
-		sample->sensor[0] = !o2pressure_sensor;
-		sample->sensor[1] = o2pressure_sensor;
-	}
-	cur_sample = sample;
-	next_o2_sensor = 0;
-}
-
-static void sample_end(void)
-{
-	if (!cur_dive)
-		return;
-
-	finish_sample(get_dc());
-	cur_sample = NULL;
-}
-
-static void divecomputer_start(void)
-{
-	struct divecomputer *dc;
-
-	/* Start from the previous dive computer */
-	dc = &cur_dive->dc;
-	while (dc->next)
-		dc = dc->next;
-
-	/* Did we already fill that in? */
-	if (dc->samples || dc->model || dc->when) {
-		struct divecomputer *newdc = calloc(1, sizeof(*newdc));
-		if (newdc) {
-			dc->next = newdc;
-			dc = newdc;
-		}
-	}
-
-	/* .. this is the one we'll use */
-	cur_dc = dc;
-	reset_dc_info(dc);
-}
-
-static void divecomputer_end(void)
-{
-	if (!cur_dc->when)
-		cur_dc->when = cur_dive->when;
-	cur_dc = NULL;
-}
-
-static void userid_start(void)
-{
-	in_userid = true;
-	//if the xml contains userid, keep saving it.
-	// don't call the prefs method here as we don't wanna
-	// actually change the preferences, this is temporary and
-	// will be reverted when the file finishes.
-
-	prefs.save_userid_local = true;
-}
-
-static void userid_stop(void)
-{
-	in_userid = false;
-}
-
 static bool entry(const char *name, char *buf)
 {
 	if (!strncmp(name, "version.program", sizeof("version.program") - 1) ||
diff --git a/core/parse.c b/core/parse.c
new file mode 100644
index 000000000..ba440f1b9
--- /dev/null
+++ b/core/parse.c
@@ -0,0 +1,449 @@
+#include <stdio.h>
+#include <assert.h>
+#include "membuffer.h"
+#include <stdlib.h>
+#include <unistd.h>
+#include <libdivecomputer/parser.h>
+
+#include "dive.h"
+#include "parse.h"
+#include "divelist.h"
+#include "device.h"
+
+/*
+static union {
+	struct event event;
+	char allocation[sizeof(struct event)+MAX_EVENT_NAME];
+} event_allocation = { .event.deleted = 1 };
+
+static event_allocation cur_event = { .event.deleted = 1 };
+static inline union {
+	struct event event;
+	char allocation[sizeof(struct event)+MAX_EVENT_NAME];
+} event_allocation = { .event.deleted = 1 };
+#define cur_event event_allocation.event
+*/
+event_allocation_t event_allocation = { .event.deleted = 1 };
+
+
+struct divecomputer *cur_dc = NULL;
+struct dive *cur_dive = NULL;
+struct dive_site *cur_dive_site = NULL;
+degrees_t cur_latitude, cur_longitude;
+dive_trip_t *cur_trip = NULL;
+struct sample *cur_sample = NULL;
+struct picture *cur_picture = NULL;
+
+bool in_settings = false;
+bool in_userid = false;
+struct tm cur_tm;
+int cur_cylinder_index, cur_ws_index;
+int lastcylinderindex, next_o2_sensor;
+int o2pressure_sensor;
+struct extra_data cur_extra_data;
+
+struct dive_table dive_table;
+struct dive_table *target_table = NULL;
+
+/*
+ * If we don't have an explicit dive computer,
+ * we use the implicit one that every dive has..
+ */
+struct divecomputer *get_dc(void)
+{
+	return cur_dc ?: &cur_dive->dc;
+}
+
+
+/* Trim a character string by removing leading and trailing white space characters.
+ * Parameter: a pointer to a null-terminated character string (buffer);
+ * Return value: length of the trimmed string, excluding the terminal 0x0 byte
+ * The original pointer (buffer) remains valid after this function has been called
+ * and points to the trimmed string */
+int trimspace(char *buffer)
+{
+	int i, size, start, end;
+	size = strlen(buffer);
+
+	if (!size)
+		return 0;
+	for(start = 0; isspace(buffer[start]); start++)
+		if (start >= size) return 0;	// Find 1st character following leading whitespace
+	for(end = size - 1; isspace(buffer[end]); end--) // Find last character before trailing whitespace
+		if (end <= 0) return 0;
+	for(i = start; i <= end; i++)		// Move the nonspace characters to the start of the string
+		buffer[i-start] = buffer[i];
+	size = end - start + 1;
+	buffer[size] = 0x0;			// then terminate the string
+	return size;				// return string length
+}
+
+/*
+ * Clear a dive_table
+ */
+void clear_table(struct dive_table *table)
+{
+	for (int i = 0; i < table->nr; i++)
+		free(table->dives[i]);
+	table->nr = 0;
+}
+
+/*
+ * Add a dive into the dive_table array
+ */
+void record_dive_to_table(struct dive *dive, struct dive_table *table)
+{
+	assert(table != NULL);
+	struct dive **dives = grow_dive_table(table);
+	int nr = table->nr;
+
+	dives[nr] = fixup_dive(dive);
+	table->nr = nr + 1;
+}
+
+void record_dive(struct dive *dive)
+{
+	record_dive_to_table(dive, &dive_table);
+}
+
+void start_match(const char *type, const char *name, char *buffer)
+{
+	if (verbose > 2)
+		printf("Matching %s '%s' (%s)\n",
+		       type, name, buffer);
+}
+
+void nonmatch(const char *type, const char *name, char *buffer)
+{
+	if (verbose > 1)
+		printf("Unable to match %s '%s' (%s)\n",
+		       type, name, buffer);
+}
+
+typedef void (*matchfn_t)(char *buffer, void *);
+
+int match(const char *pattern, int plen,
+		 const char *name,
+		 matchfn_t fn, char *buf, void *data)
+{
+	switch (name[plen]) {
+	case '\0':
+	case '.':
+		break;
+	default:
+		return 0;
+	}
+	if (memcmp(pattern, name, plen))
+		return 0;
+	fn(buf, data);
+	return 1;
+}
+
+void event_start(void)
+{
+	memset(&cur_event, 0, sizeof(cur_event));
+	cur_event.deleted = 0;	/* Active */
+}
+
+void event_end(void)
+{
+	struct divecomputer *dc = get_dc();
+	if (cur_event.type == 123) {
+		struct picture *pic = alloc_picture();
+		pic->filename = strdup(cur_event.name);
+		/* theoretically this could fail - but we didn't support multi year offsets */
+		pic->offset.seconds = cur_event.time.seconds;
+		dive_add_picture(cur_dive, pic);
+	} else {
+		struct event *ev;
+		/* At some point gas change events did not have any type. Thus we need to add
+		 * one on import, if we encounter the type one missing.
+		 */
+		if (cur_event.type == 0 && strcmp(cur_event.name, "gaschange") == 0)
+			cur_event.type = cur_event.value >> 16 > 0 ? SAMPLE_EVENT_GASCHANGE2 : SAMPLE_EVENT_GASCHANGE;
+		ev = add_event(dc, cur_event.time.seconds,
+			       cur_event.type, cur_event.flags,
+			       cur_event.value, cur_event.name);
+
+		/*
+		 * Older logs might mark the dive to be CCR by having an "SP change" event at time 0:00. Better
+		 * to mark them being CCR on import so no need for special treatments elsewhere on the code.
+		 */
+		if (ev && cur_event.time.seconds == 0 && cur_event.type == SAMPLE_EVENT_PO2 && cur_event.value && dc->divemode==OC) {
+			dc->divemode = CCR;
+		}
+
+		if (ev && event_is_gaschange(ev)) {
+			/* See try_to_fill_event() on why the filled-in index is one too big */
+			ev->gas.index = cur_event.gas.index-1;
+			if (cur_event.gas.mix.o2.permille || cur_event.gas.mix.he.permille)
+				ev->gas.mix = cur_event.gas.mix;
+		}
+	}
+	cur_event.deleted = 1;	/* No longer active */
+}
+
+/*
+ * While in some formats file boundaries are dive boundaries, in many
+ * others (as for example in our native format) there are
+ * multiple dives per file, so there can be other events too that
+ * trigger a "new dive" marker and you may get some nesting due
+ * to that. Just ignore nesting levels.
+ * On the flipside it is possible that we start an XML file that ends
+ * up having no dives in it at all - don't create a bogus empty dive
+ * for those. It's not entirely clear what is the minimum set of data
+ * to make a dive valid, but if it has no location, no date and no
+ * samples I'm pretty sure it's useless.
+ */
+bool is_dive(void)
+{
+	return (cur_dive &&
+		(cur_dive->dive_site_uuid || cur_dive->when || cur_dive->dc.samples));
+}
+
+void reset_dc_info(struct divecomputer *dc)
+{
+	/* WARN: reset dc info does't touch the dc? */
+	(void) dc;
+	lastcylinderindex = 0;
+}
+
+void reset_dc_settings(void)
+{
+	free((void *)cur_settings.dc.model);
+	free((void *)cur_settings.dc.nickname);
+	free((void *)cur_settings.dc.serial_nr);
+	free((void *)cur_settings.dc.firmware);
+	cur_settings.dc.model = NULL;
+	cur_settings.dc.nickname = NULL;
+	cur_settings.dc.serial_nr = NULL;
+	cur_settings.dc.firmware = NULL;
+	cur_settings.dc.deviceid = 0;
+}
+
+void settings_start(void)
+{
+	in_settings = true;
+}
+
+void settings_end(void)
+{
+	in_settings = false;
+}
+
+void dc_settings_start(void)
+{
+	reset_dc_settings();
+}
+
+void dc_settings_end(void)
+{
+	create_device_node(cur_settings.dc.model, cur_settings.dc.deviceid, cur_settings.dc.serial_nr,
+			   cur_settings.dc.firmware, cur_settings.dc.nickname);
+	reset_dc_settings();
+}
+
+void dive_site_start(void)
+{
+	if (cur_dive_site)
+		return;
+	cur_dive_site = calloc(1, sizeof(struct dive_site));
+}
+
+void dive_site_end(void)
+{
+	if (!cur_dive_site)
+		return;
+	if (cur_dive_site->taxonomy.nr == 0) {
+		free(cur_dive_site->taxonomy.category);
+		cur_dive_site->taxonomy.category = NULL;
+	}
+	if (cur_dive_site->uuid) {
+		struct dive_site *ds = alloc_or_get_dive_site(cur_dive_site->uuid);
+		merge_dive_site(ds, cur_dive_site);
+
+		if (verbose > 3)
+			printf("completed dive site uuid %x8 name {%s}\n", ds->uuid, ds->name);
+	}
+	free_taxonomy(&cur_dive_site->taxonomy);
+	free(cur_dive_site);
+	cur_dive_site = NULL;
+}
+
+// now we need to add the code to parse the parts of the divesite enry
+
+void dive_start(void)
+{
+	if (cur_dive)
+		return;
+	cur_dive = alloc_dive();
+	reset_dc_info(&cur_dive->dc);
+	memset(&cur_tm, 0, sizeof(cur_tm));
+	if (cur_trip) {
+		add_dive_to_trip(cur_dive, cur_trip);
+		cur_dive->tripflag = IN_TRIP;
+	}
+	o2pressure_sensor = 1;
+}
+
+void dive_end(void)
+{
+	if (!cur_dive)
+		return;
+	if (!is_dive())
+		free(cur_dive);
+	else
+		record_dive_to_table(cur_dive, target_table);
+	cur_dive = NULL;
+	cur_dc = NULL;
+	cur_latitude.udeg = 0;
+	cur_longitude.udeg = 0;
+	cur_cylinder_index = 0;
+	cur_ws_index = 0;
+}
+
+void trip_start(void)
+{
+	if (cur_trip)
+		return;
+	dive_end();
+	cur_trip = calloc(1, sizeof(dive_trip_t));
+	memset(&cur_tm, 0, sizeof(cur_tm));
+}
+
+void trip_end(void)
+{
+	if (!cur_trip)
+		return;
+	insert_trip(&cur_trip);
+	cur_trip = NULL;
+}
+
+void picture_start(void)
+{
+	cur_picture = alloc_picture();
+}
+
+void picture_end(void)
+{
+	dive_add_picture(cur_dive, cur_picture);
+	cur_picture = NULL;
+}
+
+void cylinder_start(void)
+{
+}
+
+void cylinder_end(void)
+{
+	cur_cylinder_index++;
+}
+
+void ws_start(void)
+{
+}
+
+void ws_end(void)
+{
+	cur_ws_index++;
+}
+
+/*
+ * By default the sample data does not change unless the
+ * save-file gives an explicit new value. So we copy the
+ * data from the previous sample if one exists, and then
+ * the parsing will update it as necessary.
+ *
+ * There are a few exceptions, like the sample pressure:
+ * missing sample pressure doesn't mean "same as last
+ * time", but "interpolate". We clear those ones
+ * explicitly.
+ *
+ * NOTE! We default sensor use to 0, 1 respetively for
+ * the two sensors, but for CCR dives with explicit
+ * OXYGEN bottles we set the secondary sensor to that.
+ * Then the primary sensor will be either the first
+ * or the second cylinder depending on what isn't an
+ * oxygen cylinder.
+ */
+void sample_start(void)
+{
+	struct divecomputer *dc = get_dc();
+	struct sample *sample = prepare_sample(dc);
+
+	if (sample != dc->sample) {
+		memcpy(sample, sample-1, sizeof(struct sample));
+		sample->pressure[0].mbar = 0;
+		sample->pressure[1].mbar = 0;
+	} else {
+		sample->sensor[0] = !o2pressure_sensor;
+		sample->sensor[1] = o2pressure_sensor;
+	}
+	cur_sample = sample;
+	next_o2_sensor = 0;
+}
+
+void sample_end(void)
+{
+	if (!cur_dive)
+		return;
+
+	finish_sample(get_dc());
+	cur_sample = NULL;
+}
+
+void divecomputer_start(void)
+{
+	struct divecomputer *dc;
+
+	/* Start from the previous dive computer */
+	dc = &cur_dive->dc;
+	while (dc->next)
+		dc = dc->next;
+
+	/* Did we already fill that in? */
+	if (dc->samples || dc->model || dc->when) {
+		struct divecomputer *newdc = calloc(1, sizeof(*newdc));
+		if (newdc) {
+			dc->next = newdc;
+			dc = newdc;
+		}
+	}
+
+	/* .. this is the one we'll use */
+	cur_dc = dc;
+	reset_dc_info(dc);
+}
+
+void divecomputer_end(void)
+{
+	if (!cur_dc->when)
+		cur_dc->when = cur_dive->when;
+	cur_dc = NULL;
+}
+
+void userid_start(void)
+{
+	in_userid = true;
+	//if the xml contains userid, keep saving it.
+	// don't call the prefs method here as we don't wanna
+	// actually change the preferences, this is temporary and
+	// will be reverted when the file finishes.
+
+	prefs.save_userid_local = true;
+}
+
+void userid_stop(void)
+{
+	in_userid = false;
+}
+
+void utf8_string(char *buffer, void *_res)
+{
+	char **res = _res;
+	int size;
+	size = trimspace(buffer);
+	if(size)
+		*res = strdup(buffer);
+}
+
diff --git a/core/parse.h b/core/parse.h
new file mode 100644
index 000000000..003ef9929
--- /dev/null
+++ b/core/parse.h
@@ -0,0 +1,96 @@
+// SPDX-License-Identifier: GPL-2.0
+#ifndef PARSE_H
+#define PARSE_H
+
+#define MAX_EVENT_NAME 128
+
+typedef union {
+	struct event event;
+	char allocation[sizeof(struct event) + MAX_EVENT_NAME];
+} event_allocation_t;
+
+extern event_allocation_t event_allocation;
+#define cur_event event_allocation.event
+
+/*
+ * Dive info as it is being built up..
+ */
+extern struct divecomputer *cur_dc;
+extern struct dive *cur_dive;
+extern struct dive_site *cur_dive_site;
+extern degrees_t cur_latitude, cur_longitude;
+extern dive_trip_t *cur_trip;
+extern struct sample *cur_sample;
+extern struct picture *cur_picture;
+
+
+struct {
+	struct {
+		const char *model;
+		uint32_t deviceid;
+		const char *nickname, *serial_nr, *firmware;
+	} dc;
+} cur_settings;
+
+extern bool in_settings;
+extern bool in_userid;
+extern struct tm cur_tm;
+extern int cur_cylinder_index, cur_ws_index;
+extern int lastcylinderindex, next_o2_sensor;
+extern int o2pressure_sensor;
+extern struct extra_data cur_extra_data;
+
+enum import_source {
+	UNKNOWN,
+	LIBDIVECOMPUTER,
+	DIVINGLOG,
+	UDDF,
+	SSRF_WS,
+} import_source;
+
+/* the dive table holds the overall dive list; target table points at
+ * the table we are currently filling */
+extern struct dive_table dive_table;
+extern struct dive_table *target_table;
+
+int trimspace(char *buffer);
+void clear_table(struct dive_table *table);
+void record_dive_to_table(struct dive *dive, struct dive_table *table);
+void record_dive(struct dive *dive);
+void start_match(const char *type, const char *name, char *buffer);
+void nonmatch(const char *type, const char *name, char *buffer);
+typedef void (*matchfn_t)(char *buffer, void *);
+int match(const char *pattern, int plen, const char *name, matchfn_t fn, char *buf, void *data);
+void event_start(void);
+void event_end(void);
+struct divecomputer *get_dc(void);
+
+bool is_dive(void);
+void reset_dc_info(struct divecomputer *dc);
+void reset_dc_settings(void);
+void settings_start(void);
+void settings_end(void);
+void dc_settings_start(void);
+void dc_settings_end(void);
+void dive_site_start(void);
+void dive_site_end(void);
+void dive_start(void);
+void dive_end(void);
+void trip_start(void);
+void trip_end(void);
+void picture_start(void);
+void picture_end(void);
+void cylinder_start(void);
+void cylinder_end(void);
+void ws_start(void);
+void ws_end(void);
+
+void sample_start(void);
+void sample_end(void);
+void divecomputer_start(void);
+void divecomputer_end(void);
+void userid_start(void);
+void userid_stop(void);
+void utf8_string(char *buffer, void *_res);
+
+#endif
diff --git a/packaging/ios/Subsurface-mobile/Subsurface-mobile.pro b/packaging/ios/Subsurface-mobile/Subsurface-mobile.pro
index 8332f3759..836057c67 100644
--- a/packaging/ios/Subsurface-mobile/Subsurface-mobile.pro
+++ b/packaging/ios/Subsurface-mobile/Subsurface-mobile.pro
@@ -46,6 +46,7 @@ SOURCES += ../../../subsurface-mobile-main.cpp \
     ../../../core/liquivision.c \
     ../../../core/load-git.c \
     ../../../core/parse-xml.c \
+    ../../../core/parse.c \
     ../../../core/save-html.c \
     ../../../core/statistics.c \
     ../../../core/worldmap-save.c \
diff --git a/tests/testparse.cpp b/tests/testparse.cpp
index b614560b5..49d98f66e 100644
--- a/tests/testparse.cpp
+++ b/tests/testparse.cpp
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 #include "testparse.h"
 #include "core/dive.h"
+#include "core/parse.h"
 #include "core/file.h"
 #include "core/divelist.h"
 #include <QTextStream>