src/hg/encode/encodePatchTdb/encodePatchTdb.c 1.3
1.3 2010/01/05 04:59:58 kent
Making it ignore parent track in patch.ra. Also making subTrack point to view instead of parent. Making indentation work.
Index: src/hg/encode/encodePatchTdb/encodePatchTdb.c
===================================================================
RCS file: /projects/compbio/cvsroot/kent/src/hg/encode/encodePatchTdb/encodePatchTdb.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -b -B -U 4 -r1.2 -r1.3
--- src/hg/encode/encodePatchTdb/encodePatchTdb.c 4 Jan 2010 19:12:21 -0000 1.2
+++ src/hg/encode/encodePatchTdb/encodePatchTdb.c 5 Jan 2010 04:59:58 -0000 1.3
@@ -27,14 +26,11 @@
"Example:\n"
" encodePatchTdb 849/out/trackDb.ra ~/kent/src/makeDb/trackDb/human/hg18/trackDb.wgEncode.ra\n"
"options:\n"
" -mode=mode (default %s). Operate in one of the following modes\n"
- " update - if record is new add it at end. If record is old add any new fields at end,\n"
- " and replace any old field values with new ones\n"
" replace - replace existing records rather than doing field by field update.\n"
" add - add new records at end of parent's subtrack list. Complain if record isn't new\n"
" warn if it's a new track rather than just new subtracks\n"
- " addTrack - add new track plus subtracks. Complains if not new\n"
" -test=patchFile - rather than doing patches in place, write patched output to this file\n"
" -root=/path/to/trackDb/root/dir - Sets the root directory of the trackDb.ra directory\n"
" hierarchy to be given path. By default this is ~/kent/src/hg/makeDb/trackDb.\n"
" -org=organism - try to put this at the organism level of the hierarchy instead of bottom\n"
@@ -48,8 +44,10 @@
{"root", OPTION_STRING},
{NULL, 0},
};
+boolean glReplace; // If TRUE then do a replacement operation.
+
struct loadInfo
/* Information from a stanza of a load.ra file. */
{
struct loadInfo *next;
@@ -469,8 +467,87 @@
sub->next = recordBefore->next;
recordBefore->next = sub;
}
+char *firstTagInText(char *text)
+/* Return the location of tag in text - skipping blank and comment lines and white-space */
+{
+char *s = text;
+for (;;)
+ {
+ s = skipLeadingSpaces(s);
+ if (s[0] == '#')
+ {
+ s = strchr(s, '\n');
+ }
+ else
+ break;
+ }
+return s;
+}
+
+void substituteParentText(struct raRecord *parent, struct raRecord *view,
+ struct raRecord *sub)
+/* Convert subtrack parent with subtrack view. */
+{
+struct raTag *t = raRecordMustFindTag(sub, "subTrack");
+struct dyString *dy = dyStringNew(0);
+char *s = firstTagInText(t->text);
+dyStringAppendN(dy, t->text, s - t->text);
+dyStringPrintf(dy, "subTrack %s", view->key);
+/* Skip over subTrack and name in original text. */
+int i;
+for (i=0; i<2; ++i)
+ {
+ s = skipLeadingSpaces(s);
+ s = skipToSpaces(s);
+ }
+if (s != NULL)
+ dyStringAppend(dy, s);
+else
+ dyStringAppendC(dy, '\n');
+t->text = dyStringCannibalize(&dy);
+}
+
+void indentTdbText(struct raRecord *r, int indentCount)
+/* Add spaces to start of all text in r. */
+{
+struct raTag *t;
+struct dyString *dy = dyStringNew(0);
+for (t = r->tagList; t != NULL; t = t->next)
+ {
+ char *s, *e;
+ dyStringClear(dy);
+ for (s = t->text; !isEmpty(s); s = e)
+ {
+ int i;
+ e = strchr(s, '\n');
+ if (e == s) // empty line, keep empty
+ {
+ dyStringAppendC(dy, '\n');
+ e += 1;
+ }
+ else
+ {
+ // Indent some extra.
+ for (i=0; i<indentCount; ++i)
+ dyStringAppendC(dy, ' ');
+ if (e == NULL)
+ {
+ dyStringAppend(dy, s);
+ }
+ else
+ {
+ dyStringAppendN(dy, s, e-s+1);
+ e += 1;
+ }
+ }
+ }
+ t->text = cloneString(dy->string);
+ }
+dyStringFree(&dy);
+}
+
void patchInSubtrack(struct raRecord *parent, struct raRecord *sub)
/* Patch sub into the correct view of parent */
{
if (hasViewSubtracks(parent))
@@ -486,17 +563,19 @@
struct raRecord *view = findRecordCompatibleWithRelease(parent->file, release, viewTrackName);
struct raRecord *oldSub = findRecordCompatibleWithRelease(parent->file, release, sub->key);
if (oldSub)
{
- if (sameString(clMode, "update") || sameString(clMode, "replace"))
+ if (glReplace)
{
uglyAbort("Unfortunately really don't know how to update or replace");
}
else
- recordAbort(sub, "record %s already exists - use mode update or replace",
+ recordAbort(sub, "record %s already exists - use mode=replace",
sub->key);
}
validateParentViewSub(parent, view, sub);
+ substituteParentText(parent, view, sub);
+ indentTdbText(sub, 4);
patchIntoEndOfView(sub, view);
}
else
{
@@ -514,203 +593,8 @@
patchInSubtrack(tdbParent, patch);
}
}
-struct raTag *findViewSubGroup(struct raRecord *r)
-/* Find tag that is one of the subGroup tags with first word view. May return NULL. */
-{
-int i;
-for (i=1; ; ++i)
- {
- char tagName[16];
- safef(tagName, sizeof(tagName), "subGroup%d", i);
- struct raTag *tag = raRecordFindTag(r, tagName);
- if (tag == NULL)
- return NULL;
- if (startsWithWord("view", tag->val))
- return tag;
- }
-}
-
-struct raRecord *makeParentPlusViewSubsFromComplexRecord(struct raRecord *complexRecord)
-/* Convert one complex parent record from the settings-by-view era to parent record plus views. */
-{
-struct lm *lm = lmInit(0);
-/* Get list of views from subGroup1 or subGroup2 tag. */
-struct raTag *viewSubGroupTag = findViewSubGroup(complexRecord);
-if (viewSubGroupTag == NULL)
- recordAbort(complexRecord, "Can't find view subGroup#");
-char *line = lmCloneString(lm, viewSubGroupTag->val);
-/* line looks something like:
- * view Views FiltTransfrags=Filtered_Transfrags Transfrags=Raw_Transfrags */
-char *viewWord = nextWord(&line);
-assert(sameString(viewWord, "view"));
-nextWord(&line); // Just skip over name to label views with
-struct slPair *viewList = NULL;
-char *thisEqThat;
-while ((thisEqThat = nextWord(&line)) != NULL)
- {
- char *eq = strchr(thisEqThat, '=');
- if (eq == NULL)
- recordAbort(complexRecord, "expecting this=that got %s in %s tag",
- eq, viewSubGroupTag->name);
- *eq = 0;
- slPairAdd(&viewList, thisEqThat, lmCloneString(lm, eq+1));
- }
-slReverse(&viewList);
-
-
-/* Get pointers to settingsByView and visibilityViewDefaults tags, and take them off of list. */
-struct raTag *settingsByViewTag = raRecordFindTag(complexRecord, "settingsByView");
-struct raTag *visibilityViewDefaultsTag = raRecordFindTag(complexRecord, "visibilityViewDefaults");
-struct raTag *tagList = NULL, *t, *next;
-for (t = complexRecord->tagList; t != NULL; t = next)
- {
- next = t->next;
- if (t != settingsByViewTag && t != visibilityViewDefaultsTag)
- slAddHead(&tagList, t);
- }
-slReverse(&tagList);
-complexRecord->tagList = tagList;
-
-/* Parse out visibilityViewDefaults. */
-struct hash *visHash = NULL;
-struct raTag *visTag = raRecordFindTag(complexRecord, "visibilityViewDefaults");
-if (visTag != NULL)
- {
- char *dupe = lmCloneString(lm, visTag->val);
- visHash = hashThisEqThatLine(dupe, complexRecord->startLineIx, FALSE);
- }
-
-/* Parse out settingsByView. */
-struct raTag *settingsTag = raRecordFindTag(complexRecord, "settingsByView");
-struct hash *settingsHash = hashNew(4);
-if (settingsTag != NULL)
- {
- char *dupe = lmCloneString(lm, settingsTag->val);
- char *line = dupe;
- char *viewName;
- while ((viewName = nextWord(&line)) != NULL)
- {
- char *settings = strchr(viewName, ':');
- if (settings == NULL)
- recordAbort(complexRecord, "missing colon in settingsByView '%s'", viewName);
- struct slPair *el, *list = NULL;
- *settings++ = 0;
- if (!slPairFind(viewList, viewName))
- recordAbort(complexRecord, "View '%s' in settingsByView is not defined in subGroup",
- viewName);
- char *words[32];
- int cnt,ix;
- cnt = chopByChar(settings,',',words,ArraySize(words));
- for (ix=0; ix<cnt; ix++)
- {
- char *name = words[ix];
- char *val = strchr(name, '=');
- if (val == NULL)
- recordAbort(complexRecord, "Missing equals in settingsByView on %s", name);
- *val++ = 0;
-
- AllocVar(el);
- el->name = cloneString(name);
- el->val = cloneString(val);
- slAddHead(&list,el);
- }
- slReverse(&list);
- hashAdd(settingsHash, viewName, list);
- }
- }
-
-#ifdef SOON
-/* Go through each view and write it, and then the children who are in that view. */
-struct slPair *view;
-for (view = viewList; view != NULL; view = view->next)
- {
- char viewTrackName[256];
- safef(viewTrackName, sizeof(viewTrackName), "%sView%s", complexRecord->key, view->name);
- fprintf(f, "\n"); /* Blank line to open view. */
- fprintf(f, " track %s\n", viewTrackName);
- char *shortLabel = lmCloneString(lm, view->val);
- subChar(shortLabel, '_', ' ');
- fprintf(f, " shortLabel %s\n", shortLabel);
- fprintf(f, " view %s\n", view->name);
- char *vis = NULL;
- if (visHash != NULL)
- vis = hashFindVal(visHash, view->name);
- if (vis != NULL)
- {
- int len = strlen(vis);
- boolean gotPlus = (lastChar(vis) == '+');
- if (gotPlus)
- len -= 1;
- char visOnly[len+1];
- memcpy(visOnly, vis, len);
- visOnly[len] = 0;
- fprintf(f, " visibility %s\n", visOnly);
- if (gotPlus)
- fprintf(f, " viewUi on\n");
- }
- fprintf(f, " subTrack %s\n", complexRecord->key);
- struct slPair *settingList = hashFindVal(settingsHash, view->name);
- struct slPair *setting;
- for (setting = settingList; setting != NULL; setting = setting->next)
- fprintf(f, " %s %s\n", setting->name, (char*)setting->val);
-
- /* Scan for children that are in this view. */
- struct raRecord *r;
- for (r = file->recordList; r != NULL; r = r->next)
- {
- struct raTag *subTrackTag = raRecordFindTag(r, "subTrack");
- if (subTrackTag != NULL)
- {
- if (startsWithWord(complexRecord->key, subTrackTag->val))
- {
- struct raTag *subGroupsTag = raRecordFindTag(r, "subGroups");
- if (subGroupsTag != NULL)
- {
- struct hash *hash = hashThisEqThatLine(subGroupsTag->val,
- r->startLineIx, FALSE);
- char *viewName = hashFindVal(hash, "view");
- if (viewName != NULL && sameString(viewName, view->name))
- {
- writeRecordAsSubOfView(r, f, viewTrackName);
- }
- hashFree(&hash);
- }
- }
- }
- }
- }
-#endif /* SOON */
-return NULL; // uglyf
-}
-
-
-void patchInTrack(struct raFile *tdbFile, struct raRecord *patchList)
-/* Move records in patch to appropriate place in file. Have to make up view subtracks. */
-{
-struct raFile *patchFile = patchList->file;
-linkUpParents(patchFile);
-struct raRecord *r;
-for (r=patchFile->recordList; r != NULL; r = r->next)
- {
- if (r->parent)
- validateParentSub(r->parent, r);
- char *release = raRecordFindTagVal(r, "release");
- struct raRecord *oldR = findRecordCompatibleWithRelease(tdbFile, release, r->key);
- if (oldR)
- {
- if (!sameString(clMode, "update") && !sameString(clMode, "replace"))
- {
- errAbort("track %s already exists in %s. Use mode=update or mode=replace",
- r->key, tdbFile->name);
- }
- uglyAbort("mode update and replace not yet implemented");
- }
- }
-uglyf("At least the subGroups in %s check\n", patchFile->name);
-uglyAbort("patchInTrack not implemented");
-}
void encodePatchTdb(char *patchFileName, char *tdbFileName)
/* encodePatchTdb - Lay a trackDb.ra file from the pipeline gently on top of the trackDb system. */
{
@@ -720,24 +604,23 @@
if (trackCount < 1)
errAbort("No tracks in %s", patchFileName);
boolean hasTrack = compositeFirst(patchList);
-if (hasTrack && sameString(clMode, "add"))
- errAbort("%s has a compositeTrack as well as subtracks. Use mode=addTrack to permit this.",
- patchFileName);
/* Find parent track name. */
char *parentName;
+struct raRecord *subList;
if (hasTrack)
{
parentName = patchList->key;
- checkSubsAreForParent(parentName, patchList->next);
+ subList = patchList->next;
}
else
{
parentName = cloneFirstWord(raRecordMustFindTagVal(patchList, "subTrack"));
- checkSubsAreForParent(parentName, patchList);
+ subList = patchList;
}
+checkSubsAreForParent(parentName, subList);
/* Load file to patch. */
struct raFile *tdbFile = raFileRead(tdbFileName);
int oldTdbCount = slCount(tdbFile->recordList);
@@ -746,30 +629,12 @@
oldTdbCount);
linkUpParents(tdbFile);
struct raRecord *tdbParent = findRecordCompatibleWithRelease(tdbFile, "alpha", parentName);
-if (hasTrack)
- {
- if (tdbParent)
- {
- if (!sameString(clMode, "update") && !sameString(clMode, "replace"))
- {
- errAbort("track %s already exists in %s. Use mode=update or mode=replace",
+if (!tdbParent)
+ errAbort("Can't find composite track %s compatible with alpha mode in %s",
parentName, tdbFileName);
- }
- uglyAbort("mode update and replace not yet implemented");
- }
- else
- {
- patchInTrack(tdbFile, patchList);
- }
- }
-else
- {
- if (!tdbParent)
- errAbort("parent track %s doesn't exist in %s", parentName, tdbFileName);
- patchInSubtracks(tdbParent, patchList);
- }
+patchInSubtracks(tdbParent, subList);
char *outName = tdbFileName;
if (clTest != NULL)
outName = clTest;
@@ -797,7 +661,13 @@
usage();
clRoot = simplifyPathToDir(optionVal("root", clRoot));
clMode = optionVal("mode", clMode);
clTest = optionVal("test", clTest);
+if (sameString(clMode, "add"))
+ glReplace = FALSE;
+else if (sameString(clMode, "replace"))
+ glReplace = TRUE;
+else
+ errAbort("unrecognized mode %s", clMode);
encodePatchTdb(argv[1], argv[2]);
return 0;
}