annotate csopraSplitInput.py @ 15:de71663fbce4 draft

"planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/a1d5502ae76ad46070a072641c73b7a51d21c82f/toos/CarboSeqSimulator commit a1d5502ae76ad46070a072641c73b7a51d21c82f-dirty"
author siwaa
date Tue, 10 Dec 2024 09:14:52 +0000
parents e455667fe22f
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
1 import pandas as pd
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
2 import os
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
3 import shutil
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
4 import math
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
5 import sys
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
6
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
7 def split_csv(input_folder, output_folder, max_chunck_size):
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
8 unit_path = input_folder + "/units.csv"
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
9 soil_path = input_folder + "/soil.csv"
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
10 crop_path = input_folder + "/crop.csv"
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
11
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
12 os.mkdir(output_folder)
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
13
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
14 dfs = pd.read_csv(soil_path)
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
15 dfc = pd.read_csv(crop_path)
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
16
14
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
17 fileS = open(soil_path, 'r')
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
18 fileC = open(crop_path, 'r')
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
19
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
20 LineS = fileS.readlines()
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
21 LineC = fileC.readlines()
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
22
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
23 headerS = LineS.pop(0)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
24 headerC = LineC.pop(0)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
25
10
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
26 dfs['ID'] = dfs['ID'].astype('int')
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
27 dfc['ID'] = dfc['ID'].astype('int')
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
28
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
29 IDs = dfs['ID'].unique()
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
30
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
31 max_ID = len(IDs)
5
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
32
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
33 step = max_chunck_size
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
34
10
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
35 nb_split = math.ceil(max_ID / max_chunck_size)
5
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
36
10
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
37 for i in range(1, nb_split + 1):
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
38
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
39 firstID = IDs[(i-1) * step]
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
40 lastID = IDs[min((i * step) - 1 , max_ID - 1)]
5
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
41
10
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
42 splitIDs = IDs[(i-1) * step : min((i * step) , max_ID)]
20886bc40659 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf/toos/CarboSeqSimulator commit b1cafd3eb61b631ee0b1f8f6c5ef2f9a0e2b4ccf-dirty"
siwaa
parents: 5
diff changeset
43
14
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
44 whereS = dfs['ID'].isin(splitIDs)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
45 whereSid = whereS[whereS].index.values
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
46 firstSid = min(whereSid)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
47 lastSid = max(whereSid)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
48
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
49 lineSsplit = LineS[firstSid:lastSid+1]
5
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
50
14
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
51 whereC = dfc['ID'].isin(splitIDs)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
52 whereCid = whereC[whereC].index.values
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
53 firstCid = min(whereCid)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
54 lastCid = max(whereCid)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
55
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
56 lineCsplit = LineC[firstCid:lastCid+1]
5
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
57
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
58 os.mkdir(os.path.join(output_folder, f"{input_folder}_{i}"))
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
59 shutil.copy(unit_path, os.path.join(output_folder, f"{input_folder}_{i}"))
14
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
60
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
61 splitFileS = open(os.path.join(output_folder, f"{input_folder}_{i}", "soil.csv"), 'w')
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
62 splitFileS.writelines(headerS)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
63 splitFileS.writelines(lineSsplit)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
64 splitFileS.close()
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
65
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
66 splitFileC = open(os.path.join(output_folder, f"{input_folder}_{i}", "crop.csv"), 'w')
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
67 splitFileC.writelines(headerC)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
68 splitFileC.writelines(lineCsplit)
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
69 splitFileC.close()
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
70
5
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
71 shutil.make_archive( os.path.join(output_folder,f"{input_folder}_{i}"), 'zip', os.path.join(output_folder, f"{input_folder}_{i}"))
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
72
14
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
73
e455667fe22f "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/494bf06621d4e5799cab9ef8384d4411a2cd4b84/toos/CarboSeqSimulator commit 494bf06621d4e5799cab9ef8384d4411a2cd4b84-dirty"
siwaa
parents: 10
diff changeset
74
5
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
75 # Example usage
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
76 input_folder = "input"
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
77 output_folder = "outputs"
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
78
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
79 input_file = sys.argv[1]
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
80 max_chunck_size = int(sys.argv[2])
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
81
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
82 shutil.unpack_archive(input_file, input_folder, 'zip')
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
83
3e7978af2ba2 "planemo upload for repository https://forgemia.inra.fr/carboseq/record-projet-carboseq/-/tree/5f761ea7dab0a6f8c978cbbe4946e16edc6d032e/toos/CarboSeqSimulator commit 5f761ea7dab0a6f8c978cbbe4946e16edc6d032e-dirty"
siwaa
parents:
diff changeset
84 split_csv(input_folder, output_folder, max_chunck_size)