markup_text_boxst.unk_v50_2 vector contains entries similar to general_ref

[Bumber64]

Some code I've reverse engineered starting at loc_14096F9A2 in v50.05:

int32_t unk_0 = -1;

if (string_D90 == "HF") //HISTORICAL_FIGURE
    unk_0 = 0;
if (string_D90 == "SITE")
    unk_0 = 1;
if (string_D90 == "ARTIFACT")
    unk_0 = 2;
if (string_D90 == "BOOK")
    unk_0 = 3;
if (string_D90 == "SR") //SUBREGION?
    unk_0 = 4;
if (string_D90 == "FL") //FEATURE_LAYER?
    unk_0 = 5;
if (string_D90 == "ENT") //ENTITY
    unk_0 = 6;
if (string_D90 == "AB") //ABSTRACT_BUILDING?
    unk_0 = 7;
if (string_D90 == "EPOP") //ENTITY_POP
    unk_0 = 8;
if (string_D90 == "ART_IMAGE")
    unk_0 = 9;
if (string_D90 == "ERA")
    unk_0 = 10;
if (string_D90 == "HEC")
    unk_0 = 11;

int32_t unk_4 = atoi(string_DB0);
int32_t unk_8 = -1;

if (unk_0 == 7 || unk_0 == 9)
{
    //some parsing
    unk_8 = atoi(string_D50);
}

if (unk_0 != -1)
{
    df::ref_thing *ptr = new df::ref_thing();
    ptr->unk_0 = unk_0;
    ptr->unk_4 = unk_4;
    ptr->unk_8 = unk_8;

    *world.status.unk_v50_1.unk_v50_2.push_back(ptr);
}

It looks similar to a general_ref. IDK if we're just misaligned, or it's a separate struct.

[ab9rf]

for reference, the corresponding location in v50.07-steam is 1409af072; the function begins at 14096e700

I don't understand what you mean when you say this structure "looks like" a general_refst. general_refst has no structure; this abstract base class has no data members, only a vtable pointer, and thus either looks like everything or like nothing.

Also my own analysis image has the structure I think you're talking about as (string, int16_t, int32_t, int32_t, int32_t, int32_t) although the first two int32_ts might instead be a pointer. This doesn't seem to me to match well with any of the general_refst subtypes, none of which contains a string member as far as I am aware.

[Bumber64]

It ("ref_thing") looked like a general_ref in that the tokens here that define unk_0 are similar to the enum values of general_ref_type. Structurally it's closer to specific_ref, however unk_4 and unk_8 are ids instead pointers.

It's since been determined that this struct is likely used by hyperlinks in the markup_text_box. (See the issue that mentions this one.) I guess I'll close this issue and use the other one for research instead.

[ab9rf]

i'm reopening this because we can get ground truth on this from the bay12 header dump and so we might as well do so

[Bumber64]

Looks like there won't be any need for that. I just confirmed that these are refs to legends pages by messing around with the text_boxes found in the entries in gview.view.child.page via the gm-editor in legends mode. Changing unk_0 changes the text color and the page it links to (provided unk_4 is still a valid ref, otherwise clicking does nothing.)

The enum is similar to legend_pagest.T_mode, but doesn't quite line up (no MAIN or MAPS,) so we need to create a separate one like:

<enum-type type-name='legends_page_ref_type'>
    <enum-item name='NONE' value='-1'/>
    <enum-item name='HF'/> HFS
    <enum-item name='SITE'/> SITES
    <enum-item name='ARTIFACT'/> ARTIFACTS
    <enum-item name='BOOK'/> BOOKS
    <enum-item name='SR'/> SUBREGIONS
    <enum-item name='FL'/> FEATURE_LAYERS
    <enum-item name='ENT'/> ENTITIES
    <enum-item name='AB'/> ABS
    <enum-item name='EPOP'/> POPULATIONS
    <enum-item name='ART_IMAGE'/> ART
    <enum-item name='ERA'/>
    <enum-item name='HEC'/>
</enum-type>

<struct-type type-name='legends_page_ref'>
    <enum name='type' base-type='int32_t' type-name='legends_page_ref_type'/>
    <int32_t name='ref1' init-value='-1'/>
    <int32_t name='ref2' init-value='-1'/> only used when type is AB or ART_IMAGE
</struct-type>

[ab9rf]

i've found that enum in bay12 headers (it's MarkupTextLinkType)

and the structure is called markup_text_linkst

these are both defined in markup_text_box.h