From: Raphaël Droz Date: Wed, 19 Dec 2012 01:36:29 +0000 (+0100) Subject: ldif: deeply reworked LDIF fields processing! X-Git-Tag: upstream/0.6.1~2^2~30 X-Git-Url: https://git.deb.at/?a=commitdiff_plain;h=2f827e0ef00d2c90e00adf976147faaa448b1651;p=pkg%2Fabook.git ldif: deeply reworked LDIF fields processing! * import: now relies on item_f*t() only when it's safe Internal field assignation used to be very wrong what made the LDIF import unreliable. This is not the case anymore * import: ability to consider additional aliases of LDIF attributes, eg: "zip" for "postalcode", "fax" for "facsimiletelephonenumber", ... * import: multiple emails support (up to 2, naturally) * import/export: added support for "description" (NOTES), "anniversary" and "ou" (GROUPS) * export: replaced the attribute named "xmozillaanyphone" by the _more standard_ "telephonenumber" Code has been heavily commented and hopefully opens the way to a better and more extensible LDIF importer. The last chunk of that patch is a bit unclear but very simple: LDIF export got rid of some uneeded indirections: there's no point in testing LDIF_ITEM_FIELDS and doing ldif_conv_table[] translations. Iterating over j until ITEM_FIELDS to grab names from ldif_field_names[j] is enough. Other than new features, effective changes to LDIF export brought by this commit are: * attribute order is changed (eg: the URL attribute will displayed after MOBILEPHONE) * no more empty ",mail:" in the "dn:" when no email exists * export now uses the "telephonenumber" attribute instead of "xmozillaanyphone" previously The LDIF import may be (undetectably) slower. That being said, the results a the below command are now far more reliable. $ diff ~/.abook/addressbook <( ./abook --convert --infile ~/.abook/addressbook --outformat ldif | \ ./abook --convert --informat ldif --outformat abook ) --- diff --git a/filter.c b/filter.c index de787f7..31b6f2d 100644 --- a/filter.c +++ b/filter.c @@ -505,47 +505,114 @@ export_file(char filtname[FILTNAME_LEN], char *filename) static void ldif_fix_string(char *str); -#define LDIF_ITEM_FIELDS 16 +/* During LDIF import we need more fields than the + ITEM_FIELDS of a *list_item. Eg: "objectclass" + to test valid records, ... + Here we extends the existing field_types enum + to define new fields indexes usable during processing. + Newly created LDIF attr names could be associated to + them using ldif_conv_table[]. */ +typedef enum { + LDIF_OBJECTCLASS = ITEM_FIELDS + 1 +} ldif_field_types; + +#define LDIF_ITEM_FIELDS LDIF_OBJECTCLASS typedef char *ldif_item[LDIF_ITEM_FIELDS]; +/* LDIF field's names *must* respect the ordering + defined by the field_types enum from database.h + This is only used to define (for export only) + abook standard field to LDIF attr name mappings */ static ldif_item ldif_field_names = { - "cn", - "mail", - "streetaddress", - "streetaddress2", - "locality", - "st", - "postalcode", - "countryname", - "homephone", - "description", - "homeurl", - "facsimiletelephonenumber", - "cellphone", - "xmozillaanyphone", - "xmozillanickname", - "objectclass", /* this must be the last entry */ + "cn", // NAME + "mail", // EMAIL + "streetaddress", // ADDRESS + "streetaddress2", // ADDRESS2 + "locality", // CITY + "st", // STATE + "postalcode", // ZIP + "countryname", // COUNTRY + "homephone", // PHONE + "telephonenumber", // WORKPHONE + "facsimiletelephonenumber", // FAX + "cellphone", // MOBILEPHONE + "xmozillanickname", // NICK + "homeurl", // URL + "description", // NOTES + "anniversary", // ANNIVERSARY + "ou", // GROUPS }; -static int ldif_conv_table[LDIF_ITEM_FIELDS] = { - NAME, /* "cn" */ - EMAIL, /* "mail" */ - ADDRESS, /* "streetaddress" */ - ADDRESS2, /* "streetaddress2" */ - CITY, /* "locality" */ - STATE, /* "st" */ - ZIP, /* "postalcode" */ - COUNTRY, /* "countryname" */ - PHONE, /* "homephone" */ - NOTES, /* "description" */ - URL, /* "homeurl" */ - FAX, /* "facsimiletelephonenumber" */ - MOBILEPHONE, /* "cellphone" */ - WORKPHONE, /* "xmozillaanyphone" */ - NICK, /* "xmozillanickname" */ - -1, /* "objectclass" */ /* this must be the last entry */ +/* Used to map LDIF attr names from input to + the abook restricted set of standard fields. */ +typedef struct { + char *key; + int index; +} ldif_available_items; + +static ldif_available_items ldif_conv_table[] = { + /* initial field names respect the field_types enum + from database.h but this is only for readability. + This ldif_item struct allow use to define multiple + LDIF field names ("attribute names") for one abook field */ + + {"cn", NAME}, // 0 + {"mail", EMAIL}, + {"streetaddress", ADDRESS}, + {"streetaddress2", ADDRESS2}, + {"locality", CITY}, + {"st", STATE}, + {"postalcode", ZIP}, + {"countryname", COUNTRY}, + {"homephone", PHONE}, + {"telephonenumber", WORKPHONE}, // workphone, according to Mozilla + {"facsimiletelephonenumber", FAX}, + {"cellphone", MOBILEPHONE}, + {"mozillanickname", NICK}, + {"homeurl", URL}, + {"description", NOTES}, + {"anniversary", ANNIVERSARY}, + {"ou", GROUPS}, // 16 + + // here comes a couple of aliases + {"mozillasecondemail", EMAIL}, + {"homecity", CITY}, + {"zip", ZIP}, + {"tel", PHONE}, + {"xmozillaanyphone", WORKPHONE}, // ever used ? + {"workphone", WORKPHONE}, + {"fax", FAX}, + {"telexnumber", FAX}, + {"mobilephone", MOBILEPHONE}, + {"mobile", MOBILEPHONE}, + {"xmozillanickname", NICK}, + {"labeledURI", URL}, + {"notes", NOTES}, + {"birthday", ANNIVERSARY}, + {"category", GROUPS}, + + /* TODO: + "sn": append to lastname ? + "surname": append to lastname ? + "givenname": prepend to firstname ? */ + + /* here starts dummy fields: + + As long as additional indexes are created + (using the above ldif_field_types), + any other LDIF attr name can be added and + used during ldif entry processing. + But obviously fields > ITEM_FIELDS (database.h) won't be + copied into the final *list_item. */ + + /* - to avoid mistake, don't use the special ITEM_FIELDS value. + - see also: http://mxr.mozilla.org/comm-central/source/mailnews/addrbook/src/nsAbLDIFService.cpp */ + + // used to check valid LDIF records: + {"objectclass", LDIF_OBJECTCLASS} }; +const int LDIF_IMPORTABLE_ITEM_FIELDS = (int)sizeof(ldif_conv_table)/sizeof(*ldif_conv_table); /* Handles multi-line strings. @@ -607,35 +674,90 @@ ldif_add_item(ldif_item li) list_item item; int i; - item = item_create(); + /* if there's no value for "objectclass" + it's probably a buggy record */ + if(!li[LDIF_OBJECTCLASS]) + goto bail_out; - for(i=0; i < LDIF_ITEM_FIELDS; i++) { - if(ldif_conv_table[i] >= 0 && li[i] && *li[i]) - item_fput(item,ldif_conv_table[i],xstrdup(li[i])); + /* just copy from our extended ldif_item to a regular + list_item, + TODO: API could be changed so db_fput_byid() is usable */ + item = item_create(); + for(i=0; i < ITEM_FIELDS; i++) { + if(li[i] && *li[i]) + item[i] = xstrdup(li[i]); } add_item2database(item); + item_free(&item); +bail_out: for(i=0; i < LDIF_ITEM_FIELDS; i++) xfree(li[i]); - item_free(&item); - } static void ldif_convert(ldif_item item, char *type, char *value) { - int i; - + /* this is the first (mandatory) attribute to expected + from a new valid LDIF record. + The previous record must be added to the database before + we can go further with the new one */ if(!strcmp(type, "dn")) { ldif_add_item(item); return; } - for(i=0; i < LDIF_ITEM_FIELDS - 1; i++) { - if(!strcasecmp(ldif_field_names[i], type) && *value) { - item_fput(item, i, xstrdup(value)); - break; + int i, index; + + for( i=0; i < LDIF_IMPORTABLE_ITEM_FIELDS; i++ ) { + + if( *value && // there's a value for the attr provided + ldif_conv_table[i].key && // there exists an ldif attr name... + !strcasecmp(ldif_conv_table[i].key, type)) { // ...matching that provided at input + + assert((i >= 0) && (i < LDIF_ITEM_FIELDS)); + // which abook field this attribute's name maps to ? + index = ldif_conv_table[i].index; + assert((index >= 0) && (index < LDIF_ITEM_FIELDS)); + + /* TODO: here must be handled multi-valued cases + (first or latest win, append or prepend values, ...) + Currently: emails are appended, for other fields the + first attribute found wins. + Eg: the value of "mobile" will be taken into + account if such a line comes before "cellphone". */ + + /* Remember: LDIF_ITEM_FIELDS > ITEM_FIELDS, + lower (common) indexes of *ldif_item map well to *list_item. + We can use item_fput()... */ + if(index < ITEM_FIELDS) { + // multiple email support, but two only will stay + // in the final *list_item + if(index == EMAIL && item_fget(item, EMAIL)) { + item_fput(item, + EMAIL, + strconcat(item_fget(item, EMAIL), ",", value, 0)); + } + else { + /* Don't override already initialized fields: + This is the rule of the "first win" */ + if(! item_fget(item, index)) + item_fput(item, index, xstrdup(value)); + } + } + + /* ... but if the ldif field's name index is higher + than what standards abook fields struct can hold, + these extra indexes must be managed manually. + This is the case of LDIF_OBJECTCLASS ("objectclass" attr) */ + else { + item[index] = xstrdup(value); + } + + // matching attr found and field filled: + // no further attr search is needed for `type` + break; } } } @@ -647,6 +769,11 @@ ldif_parse_file(FILE *handle) char *next_line = NULL; char *type, *value; int vlen; + + /* This is our extended fields holder to put the values from + successfully parsed LDIF attributes. + ldif_item item is temporary. When the end of an entry is reached, + values are copied into a regular *list_item struct, see ldif_add_item() */ ldif_item item; memset(item, 0, sizeof(item)); @@ -669,6 +796,7 @@ ldif_parse_file(FILE *handle) xfree(line); } while ( !feof(handle) ); + // force registration (= ldif_add_item()) of the last LDIF entry ldif_convert(item, "dn", ""); return 0; @@ -880,25 +1008,25 @@ ldif_export_database(FILE *out, struct db_enumerator e) if(*email) tmp = strdup_printf("cn=%s,mail=%s",db_name_get(e.item),email); + /* TODO: this may not be enough for a trully "Distinguished" name + needed by LDAP. Appending a random uuid could do the trick */ else tmp = strdup_printf("cn=%s",db_name_get(e.item)); ldif_fput_type_and_value(out, "dn", tmp); free(tmp); - for(j = 0; j < LDIF_ITEM_FIELDS; j++) { - if(ldif_conv_table[j] >= 0) { - if(ldif_conv_table[j] == EMAIL) { - if(*email) // don't dump en empty email field - ldif_fput_type_and_value(out, - ldif_field_names[j], - email); - } - else if(db_fget(e.item,ldif_conv_table[j])) + for(j = 0; j < ITEM_FIELDS; j++) { + if(j == EMAIL) { + if(*email) // don't dump an empty email field ldif_fput_type_and_value(out, - ldif_field_names[j], - db_fget(e.item, - ldif_conv_table[j])); + ldif_field_names[j], + email); + } + else if(db_fget(e.item,j)) { + ldif_fput_type_and_value(out, + ldif_field_names[j], + db_fget(e.item, j)); } }