Skip to content

Poor table scanning detection #56

Description

@kikaragyozov

Greetings,

You've made an excellent library. Unfortunately, given a PDF with the following table:

Image

It fails to recognize and parse it correctly.

The config I'm using is the following:

config = AutoFormatConfig(
    semantic_spanning_cells=True,
    verbosity=3
)

Saving the DataFrame to CSV I get the following output:

,Parameter \nReverse voltage (repetitive peak) \nReverse voltage (DC) \nContinuous forward current (Tc= 137°C),Parameter \nReverse voltage (repetitive peak) \nReverse voltage (DC) \nContinuous forward current (Tc= 137°C),Symbol \nVRM \nVR \nIF,Value \n650 \n650 \n10 *1,Unit \nV \nV \nA
0,Surge non-,"PW=10ms sinusoidal, Tvj=25°C",,38,A
1,repetitive forward,"PW=10ms sinusoidal, Tvj=150°C",IFSM,30,A
2,current,"PW=10μs square, Tvj=25°C",IFSM,150,A
3,Repetitive peak,forward current,IFRM,45 *2,A
4,Repetitive peak,"PW=10ms, Tvj=25°C",IFRM,7.2,A2s
5,i2t value,"PW=10ms, Tvj=150°C",∫i2dt,4.5,A2s
6,Total power dissipation,,PD,83 *3,W
7,Virtual Junction,temperature,Tvj,175,°C
8,Range of storage,temperature,Tstg,−40 to +175,°C

On another, much more simpler table, it recognizes a single column as two, and adds a void empty column for some reason (using default config here):

Image

,Symbol,Typical,Value,,Unit
0,a0,9.35,× 10,-1,V
1,a1,-1.12,× 10,-3,V/°C
2,b0,3.98,× 10,-2,Ω
3,b1,1.02,× 10,-4,Ω/°C
4,b2,1.08,× 10,-6,Ω/°C2

Am I doing something terribly wrong and missing some very needed and important configuration? I'm proficient in C#, but a complete newbie when it comes to python. Most of the code that was built to run the library was with the assistance of A.I.

Metadata

Metadata

Assignees

No one assigned

    Labels

    structure accuracyissue related to recognizing table structure ("format")

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions