Laboratoire ICAR » Plateforme TXM

For some texts, words are not highlighted in editions.

The IDS of those words not highlighted contain characters that broke the CSS ID syntax rules (e.g " ", "(" and more)

Discussion¶

Word IDs are built with <text identifier + number> or come from the sources.

If we forge the word ids in import modules, we must normalize/reduce text names to a text identifier, at the level of the corpus.

Three strategies:

• a) normalize/reduce characters or morphemes
• b) escape characters
• c) manage <text name>:<automatic text identifier> hash

b) suppose to escape with respect to the syntax reading the identifier: for example CSS syntax. So different escape algorithms may need to be used depending on context. See the XXX Java library to escape for a lot of different syntaxes.

c) suppose to use the hash in various contexts: eg concordance references, etc.

Solution¶

Define the most simple common compatible syntax compatible with CSS ID syntax and CQL syntax.

Do a) fix the XMLw to XML-TXM step of import modules, in the XML2Ana class:

• normalize/reduce the word ID to the CSS id syntax (= the syntax of xml:id)

Solution 2 (not done, see #2364)¶

• add a new import option "force word id generation" for corpora having already word IDs.
• add a new load option "force word id generation" for corpora having already word IDs.