Excel import

Import spreadsheets from Excel into existing fields. The import does not create fields. It writes into fields that already exist.

You open it in the sidebar under Data In- & Export, section Import, Excel.

An image column in the column mapping, with the footer below

File formats

The file is read in the browser. Large files do not block the interface while this happens.

Range detection

Moliri detects the filled range and shows it under Detected:

If there are title rows, intermediate headings or several tables on one sheet, correct the settings and click Re-read.

Column mapping

Moliri suggests a target field for each column, based on how closely the column header resembles the field name. Every column carries a badge that can be read without colour as well:

Not offered as targets: virtual fields, plan and wall fields, the subfields of address, object and geolocation fields, and the object ID. A column for the RegNr replaces the number that the event rule assigned on creation. Moliri only suggests image and file fields for columns that contain images or have exactly that name, so that Titel does not end up in Titelbild.

Content of the columns

Besides the column header, Moliri reads what is in the cells: datings, dimensions, years, dates, inventory numbers, yes/no values, web addresses, GND and Wikidata identifiers, file names, lists, categories, names and texts. It reads every row of the spreadsheet; a kind counts for the column if it makes up at least 60 % of the filled cells. Placeholders such as n. d., unknown or a dash are not counted. Moliri also compares the cells with the values already in the project’s fields. No value leaves the browser in the process, and no AI model runs.

This gives notices below the column:

An exact name match (sure) is never replaced, and no open column is mapped without a click. The notices appear a few seconds after the preview, a little later for large spreadsheets; until then you can already edit the mapping.

With the field search

If you have switched on the field search in your profile, you get further suggestions for the remaining columns. It is asked about columns that are still open after the column header and the content, or that have only a fuzzy target. The field search’s language model compares the column header with the meaning of the fields, so Object description finds the field Objektbeschreibung and Is highlight the field Höhepunkt (highlight). It does not read the cells; with sample values, the suggestions were less accurate in measurements.

Only fields that suit the kind of column are considered: a column with yes/no values is not offered a text field, one with dates gets a date field. Fields with an exact name match are left out. The model has to decide clearly for one field, otherwise it stays silent.

The preview does not wait for the model; its notices are added as soon as it has answered. The import never downloads the model itself: if it is not yet in the browser or in Transcribe, only the notices based on the content remain. Column headers and field names do not leave the computer in the process.

The three modes

For Fill in and Overwrite the import needs a match: a spreadsheet column and the field in the object it corresponds to (… corresponds to field in object), usually the inventory number. Rows that find no object are created with the option New objects, otherwise skipped.

A small diagram for each mode shows, on three sample rows, what File and Object turn into as the Result. It changes with the options.

Options

With Empty cells, Overwrite becomes a complete replacement: empty cells then also clear existing values.

Images in the spreadsheet

Moliri detects floating images that lie over a row, and images in cells (Excel’s newer image feature, shown as #VALUE! in older programs). For an image to appear as the title image in the object, map an image field to the image column. Moliri warns you if a column contains images but no image field is chosen.

How values are converted

The conversion depends on the target field. Values that cannot be assigned with certainty are not written but parked. The preview marks them with to resolve (Not a valid reference, parked for resolving instead of written). Above the column it says n to resolve.

Parked values are decided term by term under Resolve references . If the field is already filled by then, you choose there whether the resolved value replaces it.

JSON targets

A JSON field accepts a complete JSON document from a cell. Damaged JSON that ends with its own closing bracket is straightened out and carries the badge repaired. For text there are three routes, depending on how the field is built.

Label. If the field has labels and only one subfield per entry, such as a title in several languages, the column gets an extra Label setting. The cell value lands under this label, other labels stay as they are. Preselected is the Label beim Import (label on import) from the field’s setup, such as de; the column can choose another or enter its own. Without a label the cell is parked.

Keys. If the field has several subfields per entry, such as Herstellung (production) with place and date, the column chooses one or more of them under Keys (default whole field). One key takes the whole cell. Several keys take the cell split at the separator (default ,), in order; whatever is left over goes into the last one. The cell is taken literally.

If the field allows only one entry, the keys are merged into it; keys without a column stay as they are. The preview shows the entry as it will look after the import, an appended one with +. Header rows such as Herstellung Ort or Herstellung · Datierung preset the field and key already. If a column function returns an object, its keys apply instead of the selection.

Convert text to JSON. If the field has the step Convert text to JSON in its cleanup, the import accepts plain text instead of parking it. With , and the keys Ort, Datierung, Kommentar (place, date, comment), Berlin “Ost”, 1950 becomes an entry with the place Berlin “Ost”, the date 1950 and an empty comment. The preview shows the result and marks the cell with converted. A chosen label takes precedence over the step. See Cleanup .

Whatever reaches the field by none of these routes is parked. Under Resolve references such text can then be assigned to a key.

In the Fill in mode, for fields with labels, the label counts, not the whole field: a title with en gets de added, a de that is already filled stays. Key columns with Append also append to a filled field.

Links

Instead of a field, a column can have a link type as its target. In the field picker they are listed under Links, separated into authority data, addresses, files and objects. Above the column it then says Column is imported as link “…”. Separate several values in one cell with a line break or the separator you have set.

A value with exactly one target shows → and the target in the preview and is linked. Values without a target or with several are parked (n without target) and resolved under Resolve references . The resolved values of the same cell are linked straight away.

Column functions

For each column you can enter a small transformation under Function, which every cell passes through before the import. The preview shows the result immediately on real data.

Example: return wert.trim();. A function that has proved itself can be stored with Save as template. The templates are kept in the Import functions library (menu Data model, status 9), where they can be marked as the default per field, per field type or for all fields.

Built-in functions

Moliri comes with some functions of its own. They are listed in the selection next to Function, but only on fields they are written for: Dimensions → height (cm), for example, only on number fields.

Below the column, the import suggests the matching function (Import function: name, Apply). When splitting, the suggestion names the target fields that the project has and that the cells fill: for a list containing only 56x74cm, height and width, without depth. The column keeps its target field, such as Maßangabe (dimensions statement) for the text, and for each number field a copy of the column with its function is created. If the column has no target yet, the click also applies the suggestion based on content. No splitting suggestion comes if other columns of the file already go to height and width, and none for from and to if the file has year columns. A date field reads Excel day counts and 12.03.1987 by itself, a switch reads ja and nein; there the import suggests nothing.

Field conditions

The import checks the conditions of the fields twice: on the preview before it starts, and on the written objects afterwards. Nothing is blocked by this. Everything is imported, including a row that breaks a condition.

Before the import. If rows of the preview break a condition, Rows with unmet field conditions: n of m appears above the preview, with an overview per field and condition giving the number of rows, and under Affected rows each row individually. The check covers missing required values, rules, values that occur more than once in the spreadsheet, and values that already exist in the project or in the database. If the existing object is visible to you, (view object #…) takes you there. The footer says Rows with unmet conditions: n; a click jumps to the overview.

If the value of a unique field already exists, the overview offers alternatives to creating copies:

The overview lists skipped rows; Import again takes them back into the import.

After the import. Moliri checks each written object once as a whole, so also condition fields that were not in the spreadsheet. While the check runs, the window shows Checking field conditions: i of n objects. If it finds anything, the window stays open and shows the report: Objects with unmet field conditions: n of m, an overview per field and condition with the number of objects, and under Affected objects each object with its spreadsheet row, such as Row 5 #3173. Duplicates within the spreadsheet are listed there as well as duplicates of existing records.

Clicking an object closes the window and opens the object. The report stays available under Data In- & Export, section Import, as Import report (n) until the next import.

Footer

Before the start, the footer shows:

The preview shows five rows at first. You can change the number in the footer.

After the import

• If you like, the import continues in the background (Continue in background). Tasks shows its progress.
• The closing message names new links, parked values and unmet conditions. It offers Resolve (n) when values are parked, otherwise Report for the report . The window also names the parked values with Resolve (n).
• The history shows the import under Batch operations as one entry that can be undone as a whole.
• With New objects, the project’s creation rules apply: new objects may be created in another project and linked here.

Further reading

• Import Excel : the procedure
• Resolve references : deciding parked values
• OCR import : read PDFs and scans and take the result over here
• Bulk edit : post-processing, such as converting units