Resync opm keywords with upstream

.github/workflows/check_opmkeywords.yml

@@ -0,0 +1,34 @@
+name: Check OPM Keywords
+
+on:
+  schedule:
+    # Run weekly on Monday at 00:00 UTC
+    - cron: '0 0 * * 1'
+
+jobs:
+  check-opmkeywords:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v6
+
+      - name: Run opmkeywords sync script
+        run: |
+          set -euo pipefail
+          cd scripts
+          bash runmetoupdate.sh src/res2df/opmkeywords
+
+      - name: Check for changes
+        run: |
+          if [ -n "$(git status --porcelain src/res2df/opmkeywords/)" ]; then
+            echo "::error::OPM keywords are out of date!"
+            echo "The following files have changed:"
+            git status --porcelain src/res2df/opmkeywords/
+            echo ""
+            echo "Please run 'cd scripts && bash runmetoupdate.sh' locally and commit the changes."
+            git diff src/res2df/opmkeywords/
+            exit 1
+          else
+            echo "OPM keywords are up to date."
+          fi

scripts/runmetoupdate.sh

@@ -0,0 +1,80 @@
+#!/bin/bash
+#
+# Run this to download a fresh copy of E100 keyword metadata from opm-common
+
+# Default output directory (if not provided as an argument)
+OUTPUT_DIR=${1:-"./"}
+
+# Ensure the output directory exists
+mkdir -p "$OUTPUT_DIR"
+
+# Remember to sync the keyword list here with the list in common.py
+keywords="
+BRANPROP
+COMPDAT
+COMPLUMP
+COMPSEGS
+DENSITY
+EQLDIMS
+EQUIL
+FAULTS
+GRUPNET
+GRUPTREE
+NODEPROP
+PBVD
+PDVD
+PVDG
+PVDO
+PVTG
+PVTO
+PVTW
+ROCK
+RSVD
+RVVD
+SGFN
+SGOF
+SGWFN
+SLGOF
+SOF2
+SOF3
+SWFN
+SWOF
+TABDIMS
+VFPPROD
+VFPINJ
+WCONHIST
+WCONINJE
+WCONINJH
+WCONPROD
+WELOPEN
+WELSEGS
+WELSPECS
+WLIST
+WSEGAICD
+WSEGSICD
+WSEGVALV
+"
+
+for keyword in $keywords; do
+    firstletter=${keyword:0:1}
+    wget https://raw.githubusercontent.com/OPM/opm-common/master/opm/input/eclipse/share/keywords/000_Eclipse100/$firstletter/$keyword -O "$OUTPUT_DIR/$keyword"
+
+    # Assume that the presence of num_tables in some keywords is sort of a bug:
+    perl -p -i -e 's/num_tables/size/g' "$OUTPUT_DIR/$keyword"
+
+    # Fix multi-line comments in JSON files by replacing newlines with spaces
+    perl -0777 -p -i -e 's/"comment"\s*:\s*"((?:[^"\\]|\\.)*)"/"comment": "\1"/g; s/\n/ /g if /"comment":/' "$OUTPUT_DIR/$keyword"
+
+
+    # Pretty-print all json files (this is also done upstream)
+
+    if jq '.' "$OUTPUT_DIR/$keyword" > "$OUTPUT_DIR/$keyword.tmp"; then
+        rm "$OUTPUT_DIR/$keyword" && mv "$OUTPUT_DIR/$keyword.tmp" "$OUTPUT_DIR/$keyword"
+    else
+        rm "$OUTPUT_DIR/$keyword.tmp"
+    fi
+
+    git add "$OUTPUT_DIR/$keyword"
+done
+
+git status .

src/res2df/common.py

@@ -260,15 +260,18 @@ def keyworddata_to_df(
         else:
             dict_records.append(recdict)
         record_counter += 1
+    result_df: pd.DataFrame
     if df_records and dict_records:
         dict_df = pd.DataFrame(data=dict_records)
-        return pd.concat([*df_records, dict_df]).reset_index(drop=True)
+        result_df = pd.concat([*df_records, dict_df])
     elif df_records:  # trust that this is all one type?
-        return pd.concat(df_records).reset_index(drop=True)
+        result_df = pd.concat(df_records)
     elif dict_records:  # records contain lists.
-        return pd.DataFrame(data=dict_records).reset_index(drop=True)
+        result_df = pd.DataFrame(data=dict_records)
     else:
         return pd.DataFrame()
+    # Drop columns that are entirely NaN/None (e.g., unused optional keyword items)
+    return result_df.reset_index(drop=True).dropna(axis=1, how="all")
 
 
 def parse_opmio_deckrecord(

src/res2df/compdat.py

@@ -210,7 +210,7 @@ def deck2dfs(
                     )
                 )
                 if "INFO_TYPE" in rec_data and rec_data["INFO_TYPE"] == "ABS":
-                    rec_data["SEGMENT_MD"] = rec_data["SEGMENT_LENGTH"]
+                    rec_data["SEGMENT_MD"] = rec_data["LENGTH"]
                 welsegsrecords.append(rec_data)
         elif kword.name == "WLIST":
             for rec in kword:

src/res2df/opmkeywords/BRANPROP

@@ -3,6 +3,12 @@
   "sections": [
     "SCHEDULE"
   ],
+  "requires": [
+    "NETWORK"
+  ],
+  "prohibits": [
+    "GRUPNET"
+  ],
   "items": [
     {
       "name": "DOWNTREE_NODE",

src/res2df/opmkeywords/COMPDAT

@@ -5,70 +5,85 @@
   ],
   "items": [
     {
+      "item": 1,
       "name": "WELL",
       "value_type": "STRING"
     },
     {
+      "item": 2,
       "name": "I",
       "value_type": "INT",
       "default": 0
     },
     {
+      "item": 3,
       "name": "J",
       "value_type": "INT",
       "default": 0
     },
     {
+      "item": 4,
       "name": "K1",
       "value_type": "INT"
     },
     {
+      "item": 5,
       "name": "K2",
       "value_type": "INT"
     },
     {
+      "item": 6,
       "name": "STATE",
       "value_type": "STRING",
       "default": "OPEN"
     },
     {
+      "item": 7,
       "name": "SAT_TABLE",
       "value_type": "INT",
       "default": 0
     },
     {
+      "item": 8,
       "name": "CONNECTION_TRANSMISSIBILITY_FACTOR",
       "value_type": "DOUBLE",
       "dimension": "Viscosity*ReservoirVolume/Time*Pressure"
     },
     {
+      "item": 9,
       "name": "DIAMETER",
       "value_type": "DOUBLE",
       "dimension": "Length"
     },
     {
+      "item": 10,
       "name": "Kh",
       "value_type": "DOUBLE",
       "dimension": "Permeability*Length",
       "default": -1
     },
     {
+      "item": 11,
       "name": "SKIN",
       "value_type": "DOUBLE",
       "dimension": "1",
       "default": 0
     },
     {
+      "item": 12,
       "name": "D_FACTOR",
       "value_type": "DOUBLE",
-      "dimension": "1"
+      "dimension": "Time/GasSurfaceVolume",
+      "default": 0
     },
     {
+      "item": 13,
       "name": "DIR",
       "value_type": "STRING",
       "default": "Z"
     },
     {
+      "item": 14,
       "name": "PR",
       "value_type": "DOUBLE",
       "dimension": "Length"

src/res2df/opmkeywords/EQLDIMS

@@ -13,7 +13,7 @@
     {
       "name": "DEPTH_NODES_P",
       "value_type": "INT",
-      "default": 100
+      "default": 2000
     },
     {
       "name": "DEPTH_NODES_TAB",
@@ -23,12 +23,14 @@
     {
       "name": "NTTRVD",
       "value_type": "INT",
-      "default": 1
+      "default": 1,
+      "comment": "Not supported"
     },
     {
       "name": "NSTRVD",
       "value_type": "INT",
-      "default": 20
+      "default": 20,
+      "comment": "Not supported"
     }
   ]
 }

src/res2df/opmkeywords/EQUIL

@@ -7,59 +7,86 @@
     "keyword": "EQLDIMS",
     "item": "NTEQUL"
   },
-  "description": "The EQUIL item is used when equilibrationg the model. The item should consist of one record for each PVT region",
+  "description": "The EQUIL item is used when equilibrating the model. The item should consist of exactly one record for each EQUIL region",
   "items": [
     {
+      "item": 1,
       "name": "DATUM_DEPTH",
       "value_type": "DOUBLE",
       "default": 0,
       "dimension": "Length"
     },
     {
+      "item": 2,
       "name": "DATUM_PRESSURE",
       "value_type": "DOUBLE",
       "dimension": "Pressure"
     },
     {
+      "item": 3,
       "name": "OWC",
       "value_type": "DOUBLE",
       "default": 0,
       "dimension": "Length",
       "description": "The OWC item is depth of the OIL Water contact. This should ..."
     },
     {
+      "item": 4,
       "name": "PC_OWC",
       "value_type": "DOUBLE",
       "default": 0,
       "dimension": "Pressure"
     },
     {
+      "item": 5,
       "name": "GOC",
       "value_type": "DOUBLE",
       "default": 0,
       "dimension": "Length"
     },
     {
+      "item": 6,
       "name": "PC_GOC",
       "value_type": "DOUBLE",
       "default": 0,
       "dimension": "Pressure"
     },
     {
+      "item": 7,
       "name": "BLACK_OIL_INIT",
       "value_type": "INT",
       "default": 0
     },
     {
+      "item": 8,
       "name": "BLACK_OIL_INIT_WG",
       "value_type": "INT",
       "default": 0
     },
     {
+      "item": 9,
       "name": "OIP_INIT",
       "value_type": "INT",
       "default": -5,
-      "comment": "The default is a state variable"
+      "comment": "Only values less than or equal to zero supported"
+    },
+    {
+      "item": 10,
+      "name": "EQLOPT04",
+      "value_type": "INT",
+      "comment": "This item is ECLIPSE300 specific"
+    },
+    {
+      "item": 11,
+      "name": "EQLOPT5",
+      "value_type": "INT",
+      "comment": "This item is ECLIPSE300 specific"
+    },
+    {
+      "item": 12,
+      "name": "BLACK_OIL_INIT_HG",
+      "value_type": "INT",
+      "default": 0
     }
   ]
 }