BibTexNanny

BibTexNanny is a tool to check the consistency of BibTex files, fix common mistakes and generate simplified versions of a bibliography.

BibTex Parser

BibTexNanny uses biblib to parse and generate BibTex files.

The following fixes and changes should be made to biblib:

• Add BibDesk-compatibility mode for BibTex output
• Fix issues with loading bad month information
  • Can't replicate issue anymore, not sure what changed.
• Add ability to handle duplicate keys
• Prevent BibTex Parser from dropping metadata and comment lines
  • BibTexNanny internal work-around
• When names are parsed, curly braces need to be handled correctly

BibTex file consistency checker

• Find duplicates
  • Duplicate keys
    • added biblib work-around to load files with duplicate keys.
  • Duplicate paper titles
    • Grade badness of duplicate by how much of the rest matches
    • Consider cases where duplicates might be acceptable
      • Pairs of entries for presentation and paper (what is the entry type for the presentation).
        • Allow users to define entry types that should be ignored when looking for duplicate titles. This way you can for example model presentations as @misc entries and have them be ignored
      • Pre-print and published version of paper.
      • Author who actually named different papers differently (in what cases would this happen?)
      • Different editions of a book.
      • Possibly paper and extended version of it as journal article.
• Warnings for missing fields
  • Optional warning for optional fields
• Tex-Unicode conversion
  • LaTeX to Unicode conversion
    • Fix loosing curly braces
  • Unicode to BibTeX conversion
    • Check if URLs require special handling
• Warnings for bad formatting
  • Warning for non-standard entry type
  • Warning for fields whose value has no curly braces, but is not a known macro
  • Warnings for non-secured capitalisation in name field
  • Warnings for unnecessary curly braces
    • Curly braces are not only for uppercase characters but also for encoding special characters, e.g. \'{e} to get é
    • Allow user preference for wrapping characters or whole words.
    • What is the difference between single and double braces?
  • Warnings for badly formatted in page numbers
  • Find badly formatted names (author and editor fields)
    • All-caps names
    • Bad use of latex commands
    • Missing spaces between initials
    • Other bad formattings
  • Warning for all-caps texts
  • Notice bad months
  • Check if desired key format is followed (see entry key format)
• Warnings for inconsistent formatting
  • Different names for conferences (see dictionary of conference names)
  • Name formatting
    • Names or parts of names written in all caps (MICKEY MOUSE or Mickey MOUSE)
      • Identify when an all-caps name part is actually intials written without period or whitespace
    • Name initials
      • Initial written without period (Mickey D Mouse)
      • Multiple initials written without whitespace (Mickey A.B. Mouse)
      • Multiple initials written without periods or whitespace (Mickey AD Mouse)
      • Warning when first names are only initials
      • Warning when only some names of a paper are full and some have initials
  • Location names
    • Indicate when there is a country without a city
    • Indicate when there is a city without a country
    • States missing from US locations
  • Inferrable information for conferences/journals is inconsistent
• Allow limiting search to citations found in aux file

BibTex Fixer

• Infer fields from other entries
  • Basic inference functionality
  • Add more inferrable fields (see Field Inference)
  • Add functionality for mapping information across types (e.g. from proceeding to inproceedings)
• Infer full names
  • Infer full name form of initials when the full name is used elsewhere
  • Infer proper non-ASCII spelling of a name when is it used elsewhere
• Fix inconstistent fields
  • Replace conference name variations with main name (see dictionary of conference names)
  • Expand name initials to full names
    • Infer full name form of initials when the full name is used elsewhere
    • Infer proper non-ASCII spelling of a name when is it used elsewhere
  • Make locations more informative (City, [State], Country)
    • Add missing country
    • Add missing city
    • Add state (USA only)
    • Extend state initials to full state name
  • Have consistent file order
• Fix formatting
  • Replace non-ASCII characters in keys
  • Add wraps around capitalised characters in name field
    • Add option to wrap entire words instead of only the capitalised characters
  • Remove unnecessary {}-wraps
  • Fix badly formatted page numbers
  • Fix all-caps text (but not single all caps words)
    • Separate handling for names
  • Fix bad but understandable months (e.g. numbers)
  • Correct handling for escaped sequences - [ ] Escaped by curly braces - [ ] Escaped by math mode
  • Name formatting
    • Change format of name to non-ambiguous "Last, First" format
    • Fix special character formatting
      • Use consistent braces format (e.g. write {\"o} instead of \"{o})
      • Replace latex commands (e.g. replace \textasciicaron{}e with {e})
    • Fix all-caps names (MICKEY MOUSE or Mickey MOUSE)
    • Fix initials format
      • Initials must be followed by a period
      • Multiple initials must be separated by spaces
    • Test if text starts with "and"
• Rename entry keys
  • Provide a format to specify the desired key names
  • Key format might differ for different entry types.
  • Key format should consist of only ASCII characters
• Multi-bibliography merger
  • Identify entries that are the same
    • Option 1: Same key
    • Option 2: Match on major fields (e.g. name plus authors?)
  • Merge
    • Identical fields are accepted
    • Fields available in only one version are accepted
    • Fields that clash cause user prompt or trigger other fixer functions

BibTex simplifier

• Simplify conference names
  • Use dictionary of conference names
  • allow regex or sed replacement
• Simplify Names
  • Turn full first names into initials
  • Turn full middle names into initials
• Simplify Locations
  • Drop entirely
  • Drop city
  • Drop state
  • Shorten state to initials
  • Copy location to address (even though technically it is incorrect)

Auxiliary

Dictionary of conference names

• Allow full name, name variation, short name
• Names should allow for number placeholder
• How to link regularly named conferences with years where they were held in conjunction with something?
• Additional script to suggest possible name variations

Key formatting

• There might already be an open source system for standardising BibTex keys. This is also used by Zotero. Gotra check that out.

Relevant factors for key formatting

• First author last name
  • capitalised
  • lower caps
• Year
• Word from Title
  • capitalised
  • lower caps
  • all caps
• Disambiguating characters
  • lowercase a,b,c

Common formats

1. lastnameYEAR
2. LastnameYEAR
3. LastnameYEARkeyword
4. LastnameYEARdisambig
5. lastname_keyword_year
6. TITLEWORD
7. LastnameYEAR or KEYWORD

How to choose format

1. Number of hardcoded options
   • Easy to implement, little flexibility
2. RegEx
   • Easy to implement, flexible, but limited functionality (can't check other fields)
   • Actually, if you use named groups, you could use those names to trigger additional checks for them.
3. Custom format
   • Lots of work to implement, full functionality, probably quite flexible

Field Inference

• article: journal + year + volume => month
• article: journal + year + month => volume
• book: booktitle + year +volume/number => inbook: author, editor,publisher, series, edition, month, publisher
• book: booktitle + year +volume/number => incollection: editor, publisher, series, edition, month, publisher
• conference: booktitle + year => address, month, editor, organization, publisher
• inbook: title + year => address, month, editor, publisher
• incollection: booktitle + year => address, month, editor, publisher
• inproceedings: booktitle + year => address, month, editor, organization, publisher
• proceedings: booktitle + year => i**nproceedings: **address, month, editor, organization, publisher
• If proceedings title contains an index (e.g. "Proceedings of the 5th Conference on Examples") we can infer year and all other pieces of information from it.

BibTexNanny Input Parameters

Input methods

• Use Python's configparser, which allows INI-like config files

Internal processing

1. Dict
   • Straightforward, but need to keep the key strings straight
2. Custom object with lots of boolean fields
   • More design effort, but probably more flexible
   • Should have different class for each Nanny component
     • As the tasks overlap considerably, there should be a NannyConfig superclass and inherriting classes for the components.
     • Accessing config info should be done via functions, not fields, to allow custom processing of the stored information

Required states for custom variables

Consistency checker

• True (check value)
• False (don't check value)

Fixer

• True/Autofix/Auto (autofix value)
• Tryfix/Try (autofix if trivial, otherwise prompt to fix)
• Promptfix/Prompt (Prompt to fix)
• False (don't check value)

Consistency + Fixer

How information for both scripts can be given in the same config file

• Single value for both (Try and Prompt are treated as True)
• Tuple: False,Tryfix (CONSISTENCY,FIXER)
• Variables for only one of the two configs, e.g. duplicateKeys-consistency
• Different sections for giving instructions for both or just either

Simplifier

Should have separate config files.

• Blacklist: List fields that should be removed
• Whitelist List only the fields that are wanted
• Variables for conversion functions

============================================================

Interface

Good way to set parameters?

1. Argument calls
   • set list of wanted fields (if None, all are wanted)
   • Set list of unwanted fields (optional)
2. Config files
   • allows for templates
   • More complex to set up
3. Prompts during processing, asking for user decisions
   • Could also be used to auto-generate config files

External information files

LaTeX style files

• .bst: BibTex format file (difficult to parse)
• .sty: LaTeX style file (can this contain the bst info?)
• .cls: LaTeX class file (can this contain the bst info?)

LaTeX temp files

• .aux: Lists citations and labels
  • Single line to parse: \citation{citationlabel}

BibTexNanny files

• Dictionary of conference names
• Style config file
• Tool config files
  • Consistency checker config file
  • Fixer config file
  • Simplifier config file

BibTex field requirements

We need to be able to check the following aspects for fields:

• What type of entry are we looking at?
• What are the generally required and optional fields for this entry?
  • This bit can be hardcoded as it is always true for all BibTex files
  • Look up BibTex documentation to determine these values
• For a particular bibliography type, which are the required and optional fields, which fields are ignored?
  • Easy solution: Manually create a config file that lists fields as mandatory, optional and ignored
    • Requires config file design
  • Better solution: Load style files to automatically extract this kind of information.
    • Are there python tools that can load sty and cls files for us?
• Design a config file that allows users to set which info they want to drop and which they need enforced
  • List by entry type
    • Allow defining fields for more than one entry type at once
  • Define fields as mandatory, optional, unused and maybe as hidden
• Three layer approach:
  1. In-built BibTex entry definitions
  2. Config file for bibliography style requirements
  3. Config file for simplification requirements

People working on related tools

• Titus von der Malsburg
• Marten van Schijndel
  • Dave's fork