Browser

The TF package contains a command to work with your corpus in your browser. It sets up a local web server, which interacts with your web browser. Then you can view and search the corpus without programming and without internet connection.

Start up

Below, when you see app, you have to substitute it by the name of an existing TF app.

The TF browser fetches the apps and corpora (tf.about.corpora) it needs from GitHub / GitLab automatically.

tf app

tf app

More data

You can let TF use extra features:

tf app --mod=org/repo/path
tf app --mod=org/repo/path -c
tf app --mod=org/repo/path,org/repo/path

Here org, repo and path must be replaced with a GitHub user or organization, a GitHub repo, and a path within that repo.

Read more about your data life-cycle in the Data guide (tf.about.datasharing).

Custom sets

You can create custom sets of nodes, give them a name, and use those names in search templates. The TF browser can import those sets, so that you can use such queries in the browser too.

tf app --sets=filePath

• Start a TF browser for app.
• Loads custom sets from filePath.

filePath must specify a file on your local system (you may use ~ for your home directory). That file must have been written by calling writeSets(). If so,it contains a dictionary of named node sets. These names can be used in search templates, and the TF browser will use this dictionary to resolve those names. See Search.search(), the sets argument].

Jobs

Your session (aka job) will be saved in your browser, under the name app-default, or another name if you rename, duplicate, import or create new sessions.

Shut down

You can close the web server by pressing Ctrl-C in the terminal or command prompt where you have started tf.

Work with exported results

You can export your results to CSV files which you can process with various tools, including your own.

You can use the "Export" tab to tell the story behind your query and then export your view. A new page will open, which you can save as a PDF.

There is also a button to download all your results as data files.

Exported materials

job.json

A JSON file with all information associated with your current session. You can import this in the Jobs section, and restore the session by which you created these results.

about.md

a Markdown file with your description and some provenance metadata.

resultsx.tsv

contains your precise search results, decorated with the features you have used in your search template. Not only the results on the current page, but all results.

results.tsv

contains your precise search results, as a list of tuples of nodes. Not only the results on the current page, but all results.

sections.tsv

contains the sections you have selected as a list of nodes.

nodes.tsv

contains the nodes you have selected as a list of tuples of nodes.

nodesx.tsv

contains the nodes you have selected as a list of tuples of nodes, decorated with location information and text content.

Now, if you want to share your results for checking and replication, put all this in a research repository or in a GitHub / GitLab repository, which you can then archive to ZENODO to obtain a DOI.

UNICODE in Excel CSVs

The file resultsx.tsv is not in the usual utf8 encoding, but in utf_16 encoding. The reason is that this is the only encoding in which Excel handles CSV files properly.

So if you work with this file in Python, specify the encoding utf_16.

with open('resultsx.tsv', encoding='utf_16') as fh:
    for row in fh:
        # do something with row 

Conversely, if you want to write a CSV with Hebrew in it, to be opened in Excel, take care to:

• give the file name extension .tsv (not .csv)
• make the file tab separated (do not use the comma or semicolon!)
• use the encoding utf_16_le (not merely utf_16, nor utf8!)
• start the file with a BOM mark.

with open('mydata.tsv', 'w', encoding='utf_16_le') as fh:
    fh.write('\uFEFF')
    for row in myData:
        fh.write('\t'.join(row))
        fh.write('\n')