]> git.tdb.fi Git - ext/subsurface.git/blobdiff - parse-xml.c
Do better cylinder information management
[ext/subsurface.git] / parse-xml.c
index 5ac36a1521b5b54179308de230b23e5f5a7956a0..635bf56f208b24730761c8e60143dc54e064bd4a 100644 (file)
@@ -29,7 +29,7 @@ static void record_dive(struct dive *dive)
                dive_table.dives = dives;
                dive_table.allocated = allocated;
        }
-       dives[nr] = dive;
+       dives[nr] = fixup_dive(dive);
        dive_table.nr = nr+1;
 }
 
@@ -62,6 +62,28 @@ static int match(const char *pattern, int plen,
        return 1;
 }
 
+/*
+ * We keep our internal data in well-specified units, but
+ * the input may come in some random format. This keeps track
+ * of the incoming units.
+ */
+static struct units {
+       enum { METERS, FEET } length;
+       enum { LITER, CUFT } volume;
+       enum { BAR, PSI } pressure;
+       enum { CELSIUS, FAHRENHEIT } temperature;
+       enum { KG, LBS } weight;
+} units;
+
+/* We're going to default to SI units for input */
+static const struct units SI_units = {
+       .length = METERS,
+       .volume = LITER,
+       .pressure = BAR,
+       .temperature = CELSIUS,
+       .weight = KG
+};
+
 /*
  * Dive info as it is being built up..
  */
@@ -69,6 +91,8 @@ static int alloc_samples;
 static struct dive *dive;
 static struct sample *sample;
 static struct tm tm;
+static int suunto, uemis;
+static int event_index, cylinder_index;
 
 static time_t utc_mktime(struct tm *tm)
 {
@@ -104,14 +128,25 @@ static void divedate(char *buffer, void *_when)
 {
        int d,m,y;
        time_t *when = _when;
+       int success = 0;
 
+       success = tm.tm_sec | tm.tm_min | tm.tm_hour;
        if (sscanf(buffer, "%d.%d.%d", &d, &m, &y) == 3) {
                tm.tm_year = y;
                tm.tm_mon = m-1;
                tm.tm_mday = d;
-               if (tm.tm_sec | tm.tm_min | tm.tm_hour)
-                       *when = utc_mktime(&tm);
+       } else if (sscanf(buffer, "%d-%d-%d", &y, &m, &d) == 3) {
+               tm.tm_year = y;
+               tm.tm_mon = m-1;
+               tm.tm_mday = d;
+       } else {
+               fprintf(stderr, "Unable to parse date '%s'\n", buffer);
+               success = 0;
        }
+
+       if (success)
+               *when = utc_mktime(&tm);
+
        free(buffer);
 }
 
@@ -151,13 +186,11 @@ static void divedatetime(char *buffer, void *_when)
 }
 
 union int_or_float {
-       long i;
        double fp;
 };
 
 enum number_type {
        NEITHER,
-       INTEGER,
        FLOAT
 };
 
@@ -182,38 +215,37 @@ static enum number_type integer_or_float(char *buffer, union int_or_float *res)
                }
        }
 
-       res->i = val;
-       return INTEGER;
+       res->fp = val;
+       return FLOAT;
 }
 
 static void pressure(char *buffer, void *_press)
 {
+       double mbar;
        pressure_t *pressure = _press;
        union int_or_float val;
 
        switch (integer_or_float(buffer, &val)) {
        case FLOAT:
-               /* Maybe it's in Bar? */
-               if (val.fp < 500.0) {
-                       pressure->mbar = val.fp * 1000;
+               /* Just ignore zero values */
+               if (!val.fp)
+                       break;
+               switch (units.pressure) {
+               case BAR:
+                       /* Assume mbar, but if it's really small, it's bar */
+                       mbar = val.fp;
+                       if (mbar < 5000)
+                               mbar = mbar * 1000;
+                       break;
+               case PSI:
+                       mbar = val.fp * 68.95;
                        break;
                }
-               printf("Unknown fractional pressure reading %s\n", buffer);
-               break;
-
-       case INTEGER:
-               /*
-                * Random integer? Maybe in PSI? Or millibar already?
-                *
-                * We assume that 5 bar is a ridiculous tank pressure,
-                * so if it's smaller than 5000, it's in PSI..
-                */
-               if (val.i < 5000) {
-                       pressure->mbar = val.i * 68.95;
+               if (mbar > 5 && mbar < 500000) {
+                       pressure->mbar = mbar + 0.5;
                        break;
                }
-               pressure->mbar = val.i;
-               break;
+       /* fallthrough */
        default:
                printf("Strange pressure reading %s\n", buffer);
        }
@@ -226,12 +258,15 @@ static void depth(char *buffer, void *_depth)
        union int_or_float val;
 
        switch (integer_or_float(buffer, &val)) {
-       /* All values are probably in meters */
-       case INTEGER:
-               val.fp = val.i;
-               /* fallthrough */
        case FLOAT:
-               depth->mm = val.fp * 1000;
+               switch (units.length) {
+               case METERS:
+                       depth->mm = val.fp * 1000 + 0.5;
+                       break;
+               case FEET:
+                       depth->mm = val.fp * 304.8 + 0.5;
+                       break;
+               }
                break;
        default:
                printf("Strange depth reading %s\n", buffer);
@@ -245,28 +280,19 @@ static void temperature(char *buffer, void *_temperature)
        union int_or_float val;
 
        switch (integer_or_float(buffer, &val)) {
-       /* C or F? Who knows? Let's default to Celsius */
-       case INTEGER:
-               val.fp = val.i;
-               /* Fallthrough */
        case FLOAT:
                /* Ignore zero. It means "none" */
                if (!val.fp)
                        break;
                /* Celsius */
-               if (val.fp < 50.0) {
-                       temperature->mkelvin = (val.fp + 273.16) * 1000;
+               switch (units.temperature) {
+               case CELSIUS:
+                       temperature->mkelvin = (val.fp + 273.15) * 1000 + 0.5;
                        break;
-               }
-               /* Fahrenheit */
-               if (val.fp < 212.0) {
+               case FAHRENHEIT:
                        temperature->mkelvin = (val.fp + 459.67) * 5000/9;
                        break;
                }
-               /* Kelvin or already millikelvin */
-               if (val.fp < 1000.0)
-                       val.fp *= 1000;
-               temperature->mkelvin = val.fp;
                break;
        default:
                printf("Strange temperature reading %s\n", buffer);
@@ -300,25 +326,161 @@ static void duration(char *buffer, void *_time)
        sampletime(buffer, _time);
 }
 
-static void ignore(char *buffer, void *_time)
+static void percent(char *buffer, void *_fraction)
+{
+       fraction_t *fraction = _fraction;
+       union int_or_float val;
+
+       switch (integer_or_float(buffer, &val)) {
+       case FLOAT:
+               if (val.fp <= 100.0)
+                       fraction->permille = val.fp * 10 + 0.5;
+               break;
+
+       default:
+               printf("Strange percentage reading %s\n", buffer);
+               break;
+       }
+       free(buffer);
+}
+
+static void gasmix(char *buffer, void *_fraction)
 {
+       /* libdivecomputer does negative percentages. */
+       if (*buffer == '-')
+               return;
+       if (cylinder_index < MAX_CYLINDERS)
+               percent(buffer, _fraction);
+}
+
+static void gasmix_nitrogen(char *buffer, void *_gasmix)
+{
+       /* Ignore n2 percentages. There's no value in them. */
+}
+
+static void cylindersize(char *buffer, void *_volume)
+{
+       volume_t *volume = _volume;
+       union int_or_float val;
+
+       switch (integer_or_float(buffer, &val)) {
+       case FLOAT:
+               volume->mliter = val.fp * 1000 + 0.5;
+               break;
+
+       default:
+               printf("Strange volume reading %s\n", buffer);
+               break;
+       }
+       free(buffer);
+}
+
+static void utf8_string(char *buffer, void *_res)
+{
+       *(char **)_res = buffer;
+}
+
+/*
+ * Uemis water_pressure. In centibar. And when converting to
+ * depth, I'm just going to always use saltwater, because I
+ * think "true depth" is just stupid. From a diving standpoint,
+ * "true depth" is pretty much completely pointless, unless
+ * you're doing some kind of underwater surveying work.
+ *
+ * So I give water depths in "pressure depth", always assuming
+ * salt water. So one atmosphere per 10m.
+ */
+static void water_pressure(char *buffer, void *_depth)
+{
+       depth_t *depth = _depth;
+        union int_or_float val;
+       double atm, cm;
+
+        switch (integer_or_float(buffer, &val)) {
+        case FLOAT:
+               if (!val.fp)
+                       break;
+               /* cbar to atm */
+               atm = (val.fp / 100) / 1.01325;
+               /*
+                * atm to cm. Why not mm? The precision just isn't
+                * there.
+                */
+               cm = 100 * (atm - 1) + 0.5;
+               if (cm > 0) {
+                       depth->mm = 10 * (long)cm;
+                       break;
+               }
+       default:
+               fprintf(stderr, "Strange water pressure '%s'\n", buffer);
+       }
+       free(buffer);
 }
 
 #define MATCH(pattern, fn, dest) \
        match(pattern, strlen(pattern), name, len, fn, buf, dest)
 
+static void get_index(char *buffer, void *_i)
+{
+       int *i = _i;
+       *i = atoi(buffer);
+       free(buffer);
+}
+
+static void centibar(char *buffer, void *_pressure)
+{
+       pressure_t *pressure = _pressure;
+       union int_or_float val;
+
+       switch (integer_or_float(buffer, &val)) {
+       case FLOAT:
+               pressure->mbar = val.fp * 10 + 0.5;
+               break;
+       default:
+               fprintf(stderr, "Strange centibar pressure '%s'\n", buffer);
+       }
+       free(buffer);
+}
+
+static void decicelsius(char *buffer, void *_temp)
+{
+       temperature_t *temp = _temp;
+        union int_or_float val;
+
+        switch (integer_or_float(buffer, &val)) {
+        case FLOAT:
+               temp->mkelvin = (val.fp/10 + 273.15) * 1000 + 0.5;
+               break;
+       default:
+               fprintf(stderr, "Strange julian date: %s", buffer);
+       }
+       free(buffer);
+}
+
+static int uemis_fill_sample(struct sample *sample, const char *name, int len, char *buf)
+{
+       return  MATCH(".reading.dive_time", sampletime, &sample->time) ||
+               MATCH(".reading.water_pressure", water_pressure, &sample->depth) ||
+               MATCH(".reading.active_tank", get_index, &sample->cylinderindex) ||
+               MATCH(".reading.tank_pressure", centibar, &sample->cylinderpressure) ||
+               MATCH(".reading.dive_temperature", decicelsius, &sample->temperature) ||
+               0;
+}
+
 /* We're in samples - try to convert the random xml value to something useful */
 static void try_to_fill_sample(struct sample *sample, const char *name, char *buf)
 {
        int len = strlen(name);
 
        start_match("sample", name, buf);
-       if (MATCH(".sample.pressure", pressure, &sample->tankpressure))
+       if (MATCH(".sample.pressure", pressure, &sample->cylinderpressure))
                return;
-       if (MATCH(".sample.cylpress", pressure, &sample->tankpressure))
+       if (MATCH(".sample.cylpress", pressure, &sample->cylinderpressure))
                return;
        if (MATCH(".sample.depth", depth, &sample->depth))
                return;
+       if (MATCH(".sample.temp", temperature, &sample->temperature))
+               return;
        if (MATCH(".sample.temperature", temperature, &sample->temperature))
                return;
        if (MATCH(".sample.sampletime", sampletime, &sample->time))
@@ -326,9 +488,120 @@ static void try_to_fill_sample(struct sample *sample, const char *name, char *bu
        if (MATCH(".sample.time", sampletime, &sample->time))
                return;
 
+       if (uemis) {
+               if (uemis_fill_sample(sample, name, len, buf))
+                       return;
+       }
+
        nonmatch("sample", name, buf);
 }
 
+/*
+ * Crazy suunto xml. Look at how those o2/he things match up.
+ */
+static int suunto_dive_match(struct dive *dive, const char *name, int len, char *buf)
+{
+       return  MATCH(".o2pct", percent, &dive->cylinder[0].gasmix.o2) ||
+               MATCH(".hepct_0", percent, &dive->cylinder[0].gasmix.he) ||
+               MATCH(".o2pct_2", percent, &dive->cylinder[1].gasmix.o2) ||
+               MATCH(".hepct_1", percent, &dive->cylinder[1].gasmix.he) ||
+               MATCH(".o2pct_3", percent, &dive->cylinder[2].gasmix.o2) ||
+               MATCH(".hepct_2", percent, &dive->cylinder[2].gasmix.he) ||
+               MATCH(".o2pct_4", percent, &dive->cylinder[3].gasmix.o2) ||
+               MATCH(".hepct_3", percent, &dive->cylinder[3].gasmix.he) ||
+               MATCH(".cylindersize", cylindersize, &dive->cylinder[0].type.size) ||
+               MATCH(".cylinderworkpressure", pressure, &dive->cylinder[0].type.workingpressure) ||
+               0;
+}
+
+static int buffer_value(char *buffer)
+{
+       int val = atoi(buffer);
+       free(buffer);
+       return val;
+}
+
+static void uemis_length_unit(char *buffer, void *_unused)
+{
+       units.length = buffer_value(buffer) ? FEET : METERS;
+}
+
+static void uemis_volume_unit(char *buffer, void *_unused)
+{
+       units.volume = buffer_value(buffer) ? CUFT : LITER;
+}
+
+static void uemis_pressure_unit(char *buffer, void *_unused)
+{
+#if 0
+       units.pressure = buffer_value(buffer) ? PSI : BAR;
+#endif
+}
+
+static void uemis_temperature_unit(char *buffer, void *_unused)
+{
+       units.temperature = buffer_value(buffer) ? FAHRENHEIT : CELSIUS;
+}
+
+static void uemis_weight_unit(char *buffer, void *_unused)
+{
+       units.weight = buffer_value(buffer) ? LBS : KG;
+}
+
+static void uemis_time_unit(char *buffer, void *_unused)
+{
+}
+
+static void uemis_date_unit(char *buffer, void *_unused)
+{
+}
+
+/* Modified julian day, yay! */
+static void uemis_date_time(char *buffer, void *_when)
+{
+       time_t *when = _when;
+        union int_or_float val;
+
+        switch (integer_or_float(buffer, &val)) {
+        case FLOAT:
+               *when = (val.fp - 40587.5) * 86400;
+               break;
+       default:
+               fprintf(stderr, "Strange julian date: %s", buffer);
+       }
+       free(buffer);
+}
+
+/*
+ * Uemis doesn't know time zones. You need to do them as
+ * minutes, not hours.
+ *
+ * But that's ok, we don't track timezones yet either. We
+ * just turn everything into "localtime expressed as UTC".
+ */
+static void uemis_time_zone(char *buffer, void *_when)
+{
+       time_t *when = _when;
+       signed char tz = atoi(buffer);
+
+       *when += tz * 3600;
+}
+
+static int uemis_dive_match(struct dive *dive, const char *name, int len, char *buf)
+{
+       return  MATCH(".units.length", uemis_length_unit, &units) ||
+               MATCH(".units.volume", uemis_volume_unit, &units) ||
+               MATCH(".units.pressure", uemis_pressure_unit, &units) ||
+               MATCH(".units.temperature", uemis_temperature_unit, &units) ||
+               MATCH(".units.weight", uemis_weight_unit, &units) ||
+               MATCH(".units.time", uemis_time_unit, &units) ||
+               MATCH(".units.date", uemis_date_unit, &units) ||
+               MATCH(".date_time", uemis_date_time, &dive->when) ||
+               MATCH(".time_zone", uemis_time_zone, &dive->when) ||
+               MATCH(".ambient.temperature", decicelsius, &dive->airtemp) ||
+               0;
+}
+
 /* We're in the top-level dive xml. Try to convert whatever value to a dive value */
 static void try_to_fill_dive(struct dive *dive, const char *name, char *buf)
 {
@@ -345,6 +618,8 @@ static void try_to_fill_dive(struct dive *dive, const char *name, char *buf)
                return;
        if (MATCH(".meandepth", depth, &dive->meandepth))
                return;
+       if (MATCH(".duration", duration, &dive->duration))
+               return;
        if (MATCH(".divetime", duration, &dive->duration))
                return;
        if (MATCH(".divetimesec", duration, &dive->duration))
@@ -359,22 +634,31 @@ static void try_to_fill_dive(struct dive *dive, const char *name, char *buf)
                return;
        if (MATCH(".cylinderendpressure", pressure, &dive->end_pressure))
                return;
-       if (MATCH(".divenumber", ignore, NULL))
+       if (MATCH(".location", utf8_string, &dive->location))
                return;
-       if (MATCH(".diveseries", ignore, NULL))
+       if (MATCH(".notes", utf8_string, &dive->notes))
                return;
-       if (MATCH(".number", ignore, NULL))
+
+       if (MATCH(".cylinder.size", cylindersize, &dive->cylinder[cylinder_index].type.size))
                return;
-       if (MATCH(".size", ignore, NULL))
+       if (MATCH(".cylinder.workpressure", pressure, &dive->cylinder[cylinder_index].type.workingpressure))
                return;
-       if (MATCH(".fingerprint", ignore, NULL))
+
+       if (MATCH(".o2", gasmix, &dive->cylinder[cylinder_index].gasmix.o2))
+               return;
+       if (MATCH(".n2", gasmix_nitrogen, &dive->cylinder[cylinder_index].gasmix))
+               return;
+       if (MATCH(".he", gasmix, &dive->cylinder[cylinder_index].gasmix.he))
                return;
-       nonmatch("dive", name, buf);
-}
 
-static unsigned int dive_size(int samples)
-{
-       return sizeof(struct dive) + samples*sizeof(struct sample);
+       /* Suunto XML files are some crazy sh*t. */
+       if (suunto && suunto_dive_match(dive, name, len, buf))
+               return;
+
+       if (uemis && uemis_dive_match(dive, name, len, buf))
+               return;
+
+       nonmatch("dive", name, buf);
 }
 
 /*
@@ -421,14 +705,118 @@ static char *generate_name(struct dive *dive)
        return p;
 }
 
+static void sanitize_gasmix(gasmix_t *mix)
+{
+       unsigned int o2, he;
+
+       o2 = mix->o2.permille;
+       he = mix->he.permille;
+
+       /* Regular air: leave empty */
+       if (!he) {
+               if (!o2)
+                       return;
+               /* 20.9% or 21% O2 is just air */
+               if (o2 >= 209 && o2 <= 210) {
+                       mix->o2.permille = 0;
+                       return;
+               }
+       }
+
+       /* Sane mix? */
+       if (o2 <= 1000 && he <= 1000 && o2+he <= 1000)
+               return;
+       fprintf(stderr, "Odd gasmix: %d O2 %d He\n", o2, he);
+       memset(mix, 0, sizeof(*mix));
+}
+
+/*
+ * There are two ways to give cylinder size information:
+ *  - total amount of gas in cuft (depends on working pressure and physical size)
+ *  - physical size
+ *
+ * where "physical size" is the one that actually matters and is sane.
+ *
+ * We internally use physical size only. But we save the workingpressure
+ * so that we can do the conversion if required.
+ */
+static void sanitize_cylinder_type(cylinder_type_t *type)
+{
+       /* If we have no working pressure, it had *better* be just a physical size! */
+       if (!type->workingpressure.mbar)
+               return;
+
+       /*
+        * 35l tanks? Do they exist?
+        * Assume this is a "size in cuft" thing.
+        */
+       if (type->size.mliter > 35000) {
+               double volume_of_air = type->size.mliter * 28.317;      /* cu ft to milliliter */
+               double atm = type->workingpressure.mbar / 1013.25;      /* working pressure in atm */
+               double volume = volume_of_air / atm;                    /* milliliters at 1 atm: "true size" */
+               type->size.mliter = volume;
+       }
+}
+
+static void sanitize_cylinder_info(struct dive *dive)
+{
+       int i;
+
+       for (i = 0; i < MAX_CYLINDERS; i++) {
+               sanitize_gasmix(&dive->cylinder[i].gasmix);
+               sanitize_cylinder_type(&dive->cylinder[i].type);
+       }
+}
+
 static void dive_end(void)
 {
        if (!dive)
                return;
        if (!dive->name)
                dive->name = generate_name(dive);
+       sanitize_cylinder_info(dive);
        record_dive(dive);
        dive = NULL;
+       cylinder_index = 0;
+}
+
+static void suunto_start(void)
+{
+       suunto++;
+       units = SI_units;
+}
+
+static void suunto_end(void)
+{
+       suunto--;
+}
+
+static void uemis_start(void)
+{
+       uemis++;
+       units = SI_units;
+}
+
+static void uemis_end(void)
+{
+}
+
+static void event_start(void)
+{
+}
+
+static void event_end(void)
+{
+       event_index++;
+}
+
+static void cylinder_start(void)
+{
+}
+
+static void cylinder_end(void)
+{
+       cylinder_index++;
 }
 
 static void sample_start(void)
@@ -449,6 +837,7 @@ static void sample_start(void)
        }
        sample = dive->sample + nr;
        memset(sample, 0, sizeof(*sample));
+       event_index = 0;
 }
 
 static void sample_end(void)
@@ -456,14 +845,6 @@ static void sample_end(void)
        if (!dive)
                return;
 
-       if (sample->time.seconds > dive->duration.seconds) {
-               if (sample->depth.mm)
-                       dive->duration = sample->time;
-       }
-
-       if (sample->depth.mm > dive->maxdepth.mm)
-               dive->maxdepth.mm = sample->depth.mm;
-
        sample = NULL;
        dive->samples++;
 }
@@ -564,37 +945,62 @@ static void visit(xmlNode *n)
        traverse(n->children);
 }
 
+/*
+ * I'm sure this could be done as some fancy DTD rules.
+ * It's just not worth the headache.
+ */
+static struct nesting {
+       const char *name;
+       void (*start)(void), (*end)(void);
+} nesting[] = {
+       { "dive", dive_start, dive_end },
+       { "SUUNTO", suunto_start, suunto_end },
+       { "sample", sample_start, sample_end },
+       { "SAMPLE", sample_start, sample_end },
+       { "reading", sample_start, sample_end },
+       { "event", event_start, event_end },
+       { "gasmix", cylinder_start, cylinder_end },
+       { "cylinder", cylinder_start, cylinder_end },
+       { "pre_dive", uemis_start, uemis_end },
+       { NULL, }
+};
+
 static void traverse(xmlNode *root)
 {
        xmlNode *n;
 
        for (n = root; n; n = n->next) {
-               /* XML from libdivecomputer: 'dive' per new dive */
-               if (!strcmp(n->name, "dive")) {
-                       dive_start();
-                       visit(n);
-                       dive_end();
-                       continue;
-               }
+               struct nesting *rule = nesting;
 
-               /*
-                * At least both libdivecomputer and Suunto
-                * agree on "sample".
-                *
-                * Well - almost. Ignore case.
-                */
-               if (!strcasecmp(n->name, "sample")) {
-                       sample_start();
-                       visit(n);
-                       sample_end();
-                       continue;
-               }
+               do {
+                       if (!strcmp(rule->name, n->name))
+                               break;
+                       rule++;
+               } while (rule->name);
 
-               /* Anything else - just visit it and recurse */
+               if (rule->start)
+                       rule->start();
                visit(n);
+               if (rule->end)
+                       rule->end();
        }
 }
 
+/* Per-file reset */
+static void reset_all(void)
+{
+       /*
+        * We reset the units for each file. You'd think it was
+        * a per-dive property, but I'm not going to trust people
+        * to do per-dive setup. If the xml does have per-dive
+        * data within one file, we might have to reset it per
+        * dive for that format.
+        */
+       units = SI_units;
+       suunto = 0;
+       uemis = 0;
+}
+
 void parse_xml_file(const char *filename)
 {
        xmlDoc *doc;
@@ -605,6 +1011,7 @@ void parse_xml_file(const char *filename)
                return;
        }
 
+       reset_all();
        dive_start();
        traverse(xmlDocGetRootElement(doc));
        dive_end();