File formats

RigakuFile auto-detects which of the two common Rigaku text layouts a file uses.

Canonical .ras

Canonical .ras files are plain text with section markers. A typical single-scan file looks like this:

*RAS_DATA_START
*RAS_HEADER_START
*FILE_SAMPLE "ZIF-62 Test"
*HW_XG_TARGET_NAME "Cu"
*HW_XG_WAVE_LENGTH_ALPHA1 "1.540593"
*MEAS_SCAN_AXIS_X "TwoThetaTheta"
*MEAS_SCAN_UNIT_X "deg"
*MEAS_SCAN_UNIT_Y "cps"
...
*RAS_HEADER_END
*RAS_INT_START
10.0000 100.5
10.5000 150.2
11.0000 300.8
*RAS_INT_END
*RAS_DATA_END

Multi-scan files contain several *RAS_HEADER_START … *RAS_INT_END blocks inside a single outer *RAS_DATA_START … *RAS_DATA_END wrapper; each block is parsed into its own RigakuScan element of the returned RigakuFile.

Some exports include a third column after intensity (attenuator coefficient). It is parsed and silently discarded — only the first two columns are stored.

Simplified .txt

The simplified text export drops the section markers. Metadata lines still start with *KEY "VALUE", annotations with #key=value, and the data block is just whitespace-separated columns:

*FILE_SAMPLE "ZIF-62 Test"
*HW_XG_TARGET_NAME "Cu"
*HW_XG_WAVE_LENGTH_ALPHA1 "1.540593"
*MEAS_SCAN_UNIT_X "deg"
*MEAS_SCAN_UNIT_Y "cps"
#Intensity_unit=cps
10.0000 100.5
10.5000 150.2
11.0000 300.8

Annotation lines are stored in scan.metadata under "_" * key (so #Intensity_unit=cps becomes scan.metadata["_Intensity_unit"] == "cps") to avoid colliding with regular *KEY entries.

Missing-field sentinels

To keep RigakuScan concretely typed, missing metadata uses sentinel values rather than missing:

Check haskey(scan.metadata, "HW_XG_WAVE_LENGTH_ALPHA2") when you need to distinguish a genuinely recorded zero from a missing value.

Datetime formats

MEAS_SCAN_START_TIME and MEAS_SCAN_END_TIME are parsed using these formats, in order:

1. "m/d/y H:M:S" (US style, as emitted by most Rigaku firmware)
2. "y/m/d H:M:S" (occasional alternate ordering)

Unparseable non-empty timestamps emit a warning and fall back to DateTime(1).

Encoding and line endings

Files are read as plain text via readlines. Both Unix (LF) and Windows (CRLF) line endings are supported. Files with non-ASCII characters in metadata strings (e.g. sample names in Japanese) should be UTF-8 encoded. Older Shift-JIS Rigaku exports are not currently decoded automatically — convert to UTF-8 first, e.g. with iconv -f SHIFT-JIS -t UTF-8.